Unlocking Creativity through Artificial Intelligence with Varun Gulati of Koalluh

intelligence. Varun is a former public school teacher, and a product and engineering leader. he co founded, scaled and sold his first company you class as a CTO and head of product. Most recently, Varun led the product and growth teams at Mystery Science the fastest growing K 12 Science supplemental in the US. Varun is also an ex Googler and

a former TFA corps member. We spoke with her and recently along with his co founder, Christy Chu, as they're one of the companies pitching at South by Southwest edu is upcoming launch competition. Varun Gulati, welcome to Ed Tech insiders.

But one thing that's very special about him is that he was an avid and very hungry learner. So through his own advocacy, and through a combination of people who saw that in him, he really transformed his entire life outcome by like, seeking out education and learning. And that impacted the generations to come, including me. So I grew up with my parents and my grandparents were really preaching about the value of

education and learning. And as I grew older, I started to separate the two concepts out from my starving, evangelize to me, right, like education and learning. And think about that a lot more deeply. And for myself, I've had some great learning experiences in formal education environments, but almost also some not so great ones. And I think a lot about the bet the magic that happens when formal education is more or less an enabler for a kid's curiosity to

thrive and hold. An example this just a little side note is in high school, I had to take an elective, and I chose journalism. And it was in that class that might my teacher, Mr. mouth when he showed me and my fellow classmates that journalism was really just about listening, and thinking and communicating to you. And prior to this, Alex, I hated everything to do with clinical ELA. I was I thought I was a bad reader. I thought I was a bad

writer. And so I avoided it. And he changed that entirely changed the narrative in my head, and he changed my entire orientation orientation towards that. And the lot of learning, learning that happened beyond that, for me, was enabled by him, right? I got involved in the high school newspaper, and became one of the youngest newsletters on my college. It became like a thing

that I sought out. And that's what I'm what has been really captivating me and my journey in education is how can learning be a practice of joy and curiosity and self discovery? And digressing over you asked me that?

everything. I loved working with the students, so condition, many of them, there will be some things that were challenging, like, there are many moments that I felt that I was teaching to the test or were It felt as foods were learning, quote, unquote, but they weren't experiencing the joy of learning. And so I walked away from the experience with like, all these thoughts and feelings, I had a brief stint at Google.

And then I co founded my first company, which at the time is called view class at the time, it was basically a project based learning lesson creator and exchange. So teachers could create a PBL lessons and teach them alongside other teachers all around the world. So we grew that and through a series of pivots, we found product market fit with a drop box for gauge Well, I we ended up selling the company to Renaissance Learning, that was really the Renaissance for a few years, mostly in like

a product coaching capacity. And that's it. It's towards my my tail end of my time over here that I met Keith from Mystery Science. And we really connected about a lot of things. But one thing I really loved about Mystery Science was that it brought out a lot of the things that I was talking about earlier, the joy and the curiosity of learning, and just magical deceit like kids talked about it. And to see how much teachers love your admins loved

it, and parents loved it. So started working this through science eventually joined full time. And then I most recently was leading the project, the growth teams. And then I was doing until May, and in July, my co founder, Christy and I, we went full time on koala. And we were poking around at a few different problems faces, we got very intrigued about reading engagement, and reading engagement. Excited as specifically because it's so

complicated. On one hand, yeah, it's appropriate those experience it, it's obvious like we're willing to like if you if you have someone who's engaged in reading, it can immerse you in stories to hold it from the universe's there for many others. It's that reading engagement, it's actually an accessible right now. And that's the problem we're excited to solve. We're thinking about how can we bring the joy of reading and writing to every learner, while helping them develop their

reading and writing skills? And yeah, that's what we set out to do.

And it feels like there's something very parallel to what you're trying to do with reading now. And what you've done with science and what you've done with it, like generally, it's shifting frames of reference to allow people to see themselves in the learning and make it feel more like learning than education. Like as you say, more like it's they're doing it than it's being done to them.

is on everybody's minds. I see you as ahead of the curve on this, you know, now that chat GPT and, and mid journey and all of these different creative tools have sort of caught the whole space on fire, you are already doing this, you are already thinking about how generative AI can complement

creative work for kids. So I'd love to hear how you sort of came to that conclusion and how you put together the sort of slightly hard skill of AI the technical skill and the really soft, emotional, psychological piece that comes with with reading and telling stories?

a number of different angles. We heard about it from parents, we had parents say that it felt like pulling teeth to get their kids to do 20 minutes of daily reading a day. We connected to our own personal experiences, I shared a little bit about my own challenges with how I viewed reading, and, you know, ela. Then we started to like dig into the experiences of other educators and connected to data as well. And one thing that we found is that only 17% of kids

are read for fun. And that's a stat that's plummeting every year. It used to be more than double when I was growing up. So it's it's dropping every year. And the aha moment for us came when we were invited to the data of your journey. And both Chris and I have exact same realization at the same time, which was, wouldn't it be cool to put this technology in the hands of kids so Already we've been chewing on this right on the side of like, how do we

engage kids? How do we get kids excited about, like, about about reading and writing. And then we were playing with this with these tools that were so exciting. And that's what got us down the path of like, what are some ways that we can use this as a as a creative tool to get kids in game. So we think around the three different prototypes tested with a bunch of kids and seeing, seeing the magic of kids playing with, it really confirmed for us that this was a direction we wanted to double

down on. So yeah, that's a built in after that we built in the eyes for a generator, the idea being that if you can involve kids in story creation process, if you can turn them to the co authors of stories, then you'll significantly increase motivation. And one thing that's really exciting about this is that technology is both both interesting to kids as a means and as a means to the end that it produces.

sense. And I think it is something we all sort of feel a little bit whenever you play with something like mid journey or you know, experiment with any of these tools, it feels like you're sort of co authoring something you're sort of spitting out ideas or telling the the AI what you're interested in, or what you like, or what sort of, you know, what you'd like to see. And then the AI sort of takes it further and makes it high fidelity or makes it you know, takes it adds a lot

of facts or things like that. I love that idea of co authoring. And you know, one thing that strikes me about Qualla, I'll ask this quickly is just like, I think the idea of putting AI in the hands of kids is I still think is a little bit ahead of the curve, you're seeing other you know, business models start to spring up left and right with you how to use these AI engines. But as far as I've seen so far, most of them are giving more

power to teachers. Sometimes they're giving chatbots to kids and ways for kids to sort of communicate with an AI. But it's it's often in a very sort of give and take like I need the answer to this, you can give me

a hint kind of thing. So I think you're you're doing something really interesting with AI and sort of co authoring and creativity, help our listeners and me understand, you know, how these pieces can come together in a way that really excites and motivates kids to create their own stories, and then to transfer that to engagement with reading in general?

It's amazing. I think like I mean, this is not like novel information. I mean, educators have thought about this for a long time. I mean, just to say like two examples. There's a the four C's rubric, for example, which creativity, collaboration, etc. There are different levels in which educators have thought about bringing creativity to students. And we're, I would say students have that approach ourselves, right? We're actively learning and discovering ways to incorporate that into the

product. And the second thing I mentioned is that we're very inspired by a lot of the constructivist approaches as well as their constructive, meaning like learners building their own mental models, and then using that to like, develop their understanding and incorporate new information. So an example of this is John Piaget, he he's a psychologist, studied a lot of cognitive

development. And he cited this difference between given knowledge and knowledge constructed in the classroom to we tend to separate out practice and instruction to meet buckets. But it's actually really magical to see students who are doing the knowledge construction, who are learning through play. An example of this is for just in my own life, when I was in sixth grade, I think I was I took a geometry class in high school, and I hated the proofs and the algorithms and blah, blah, blah.

drove me crazy. And then in like our computer lab portion of it, we use this tool called GeoGebra. Have you ever heard of it? Yeah. It's arguably not a great piece of software. But it is such a great example of like, a constructivist piece. Cool, because you could, yeah, you create anything. And I learned geometry through constructing my own understanding of spatial awareness. And I don't know if there's like a clean analogy to that in the reading world just yet. But that would be amazing

for us to like, become that. And so that's what we're really interested in real interested in, like, how do we bring more creative controls to kids? And there's actually a very, like, natural point of where writing can fit into this. And there's both from the research world where there's A significant amount of studies have been done, which pointed the crucial role of writing literacy development, the writing and reading goes hand in hand,

right? Like in one case, you're reading, you're working with text that's given to you in writing, you're constructing your texts. And there's, there's like a guarantee constructivist approach in that, and I think we're really interested in challenging the binary of, of that text given to a kid versus text generated by kin. All right, like kind of Nico Pilar model, we're closer to texts

given to a kid. But it's really exciting to be able to, like try different paradigms, where we can move more towards students and more towards students constructing, and more towards the writing end of it as well. So yeah, I think creativity is super exciting. We're very jazzed about like, how we can bring more creative controls to kids.

And, you know, what koala does is really allows kids to bring their curiosity, their interests, their, their, your interests from from their life, or from their previous reading, but even if they don't read, the things they just care about are interested in life, to the reading experience, and they can

actually make it happen. I think the Geogebra example is really interesting, because you know, the difference between doing a geometry worksheet where you're seeing 10 circles, or, you know, rhombuses, and having to figure something out about them, and sitting in actually making a rhombus and moving it around and moving the angles and trying different things, it's a big

difference. It's a very different way to see the same material, I think that the idea of being able to change what's happening in your reading to create new chapters, create new characters to create new settings, and then see them both written and illustrated, would probably be a pretty good mental shift for kids. And I think you've probably seen that in the classroom.

penguins. And sure enough, you wrote books about penguins. And that's, it's just so cool to see that like that there are more and more waves that are coming out, that will have that allow kids to push the boundaries in terms of play and creation. And that's how they develop their own understanding of the world. Yeah,

interests. And I love the idea of not feeling like these different types of topics have to be so different, in a kid's mind being able to bring some of that interest. Kids are obsessed with penguins, they should absolutely be able to read and write stories about penguins until they have a new obsession.

It's a great practice. And it's an even, there's even great research about interest based learning where it especially with reading, where kids who already care about a topic, they've tested this with sports, for example, kids who care about sports are literally better at reading when they're reading about sports. They literally comprehend more. And it feels like it feeds right into what you're doing with coil I, I want to ask about the classroom

environment. So you know, a koala is you have a beta on the site. Now kids can create their own stories and illustrations whenever they'd like at home. But I know that there's also a really interesting use case in a classroom environment. So how do you envision koala being used in the schools?

engagement. Right? When kids are creative, they're engaged. That's there's anecdotal evidence for that empirical evidence for that study, so on and so forth. And teachers care a lot about engagement. I cared a lot about engagement. Every almost every teacher I've talked to is like they cited as one of the top three things they care

about. And it's important to because given all the focus on signs of reading, a very critical piece of that is self regulation is specifically around like motivation and engagement as it plays into the process of learning to read. And that's one of the week that we care a lot about that teachers care a lot about in qual as a way to engage students in reading and seeing writing as well in the classroom. That's

one piece of it. Like, again, there's a lot of interesting research, I'm happy to like mention that one of our one of the people that we've been chatting with that University of California Irvine, Dr. Yong Kim, she introduces to the what's called Dar, the direct and indirect effects model of leading. And what's cool in Google dir reading on Google image search, you will see this, like Coliseum like graphic come up. And in the middle of it is social emotional development.

And social emotional development is it encompasses a few different things, but motivation engagement, are the are two of the biggest ingredients of that. And that interacts with every element, it's a read, there's a reason why it's at the middle interacts with the word reading with text, reading fluency. With text structure, you name it, this is a very critical piece of

the science of reading. The second thing I'll point out is that the number one thing we've heard from teachers that they're struggling with, in terms of reading, is differentiation, quality. Now, it's a way for students to engage in reading practice in a way that's fun and easy. But what would make it even stronger we're working towards is making it so we can support all learners at wherever

they're at. My co founder Christie, when she was teaching secondary, most recently, she had students reading at if you were to track it out to grade levels, she had students are reading at four different grade levels. And, you know, that's really challenging for a teacher to work with. And this never really been possible before to actually meet students needs add where they are, as closely as we

can. Now with the power of AI, the previously you you approximate a levels and give a tax that's close, we can get a lot closer and much more personalized at the same time. So that's the second thing that's really fascinating is like the differentiation piece. And the thing that the third thing I'll say, Alex is that we actually don't want to be too prescriptive, that's learning environments are guided by teachers. And we want to allow for a lot of freedom and exploration on the student side

as well. So there are obvious use cases like we've been used as an individual reading time tool, less obvious use cases, seizures, we've used this in small group who have created stories in front of their class guided by their class and a whole group setting. So I think the exciting thing is, with a tool like this that has a little more of the creative capacity, there, there are certain obvious

ways that you can use it. But there are also a lot of ways that teachers can use it in a way that feels right for them. It's

the role of AI. And I think there's a little bit of a real curiosity in the field, you know, they've been working for years to be able to differentiate among different grade levels grading by hand, basically, but using an expertise. And suddenly you have this AI that could at least attempt if not very successfully, do that type of grade differentiation. So you have this interest based, you know, generation, you have this potential differentiation that

could happen. And then you have the ability, as you say, to try different ways that could be communal learning where you're reading together, creating for each other, there can be individual learning, there can be a group of kids who all like the same thing, making a story, you know, together, there's so many interesting new models that are enabled when you have tools that can create reading on the fly, basically create meaningful, and hopefully, you know, accurate reading passages

on the fly either, you know, stimulated or generated by kids themselves or by teachers. Just one question to follow up on your penguin use case before. You're saying, hey, this teacher knew already coming in that her kids were that our whole class is obsessed with penguins. And that was going to be motivated to them that would cover some of the social emotional pieces, so they were going to likely create their own penguins. Have you heard teachers say, Oh, this is great because I have 10

different kids. I know they all like different things, but I don't have the time to make things that are interesting to each of them or things that meet their exact grade level. But this could allow me or them to, you know, create instant readings that are specific to each child. That is

worlds of their peers as well. I mean, a big component of this is that we allow kids to read things that the stories that other kids are creating in their classroom. There are many reasons why that's important. Learning is a social thing in real life, and it should be social in a digital realm as well. Do you

just sort of co authored. How are you doing the illustrations and sort of what do you see as the future of that kind of visual stimulation as part of reading engagement?

pictures. So our way around this is to generate pictures per paragraph so that they're not like inferring too much about what's happening in the text through by guessing. And while looking at the pictures. The point of it is that kids really like looking at illustrations, regardless of like,

instructional use case of it. So we're currently, like I said, using open source models, but one of our goals for for next quarter is to improve, improve those models of that one, their higher fidelity, right now they can be kind of hit or miss. And to to allow kids to be able to like shape the direction of those illustrations as well as they see fit.

didn't occur to me at all. I wonder if there's an interesting opportunity to sort of, for lack of a better word gamify this in that maybe students can only create illustrations after they've successfully decoded, you know, the words and then it becomes this amazing way to sort of make the whole page come alive. Has that been something that you've thought about or anybody has expressed interest in?

text and images. I'm curious if you've gotten any nervousness or pushback from teachers about the idea that this will generate content that is direct to kids, how do you make sure that it's appropriate, and every way that it's the text is appropriate, but also that the visuals are, you know, safe, basically, for kids?

freeform. And we had kids basically taking like an open text box and type in whatever they want, when we constructed a story out of that. And one thing that we found is that like there was too much variability in how the story would come out. So we tried a few different things we created as like a safety module that filtered before we generated the check And then filtered again afterwards. But ultimately what we did is for the time being, we've limited the number of options on the

public beta. And what we want to do now is improve our safety module so that we can bring that back to kids. So to put it another way, until we can guarantee with near 100% accuracy that is going to be totally safe for kids, we're not going to introduce a new creative control into the mix. Because it's important for us that teachers feel that they can trust this as of now they can. And we need to continue to maintain that trust with teachers,

potentially relevant here. And I thought could be interesting, which is that open AI, his prediction is that open AI is is going to start putting out API's that allow people to actually add constraints to what the technology can do, because you know, people have been, especially kids, you know, like to push the limits they like to ask, and journalists, students and journalists both like to sort of see where the edges are, and where how you can run off the side of the of the map, and

people have been pretty successful at it. And you can run pretty far off the map. And I think that open AI is, is very aware of that. And they're starting to think about how they can in their own models, start to you know, create opportunities for safety and make sure that things are not going to be inappropriate. And then, as you mentioned, being able to add additional safety modules and layers within it to make it surely truly school appropriate. That feels really

powerful. And it feels like something that schools would be very interested in to be able to use these amazing new tools in a way that they know is safe, and they're not going to get you know, parent phone calls, let alone, you know, make a kid believe in sentient AI or freak out in some way. So I love this idea. And I'm so curious, just because you've been thinking

about this for a while? How long do you think it will be before koala or open AI or just the EdTech industry feels really safe putting AI that is pretty open ended into the classroom environment?

about? Not just from like, is this a for as an art perspective, but also is the things that the teacher is wanting to like, guide students to read about, you know, there, and we're, again, looking at this from an age appropriateness level, from a content purpose level, but also from the lens of how do we empower the teacher to make some of these decisions as well, ultimately, the teacher is the guide, and the person who is leading the classroom, we want them to feel and know that they

can use this technology for themselves, it's for them to use is not something that like, you know, it's we want, like people to feel fear or feel scared of?

how you think about that. And if that's, you know, what that might look like in the future?

like that topic as well? And yeah, that's one of the experiments that we're designing right now is around introducing some like, capacity for educators to I don't wanna say control the to a guide, the specific session around a specific topic.

And then I want to zoom out and talk about, you know, AI and education generally I know we've already begun under some of that, this might be a little bit of a strange question. But I'm so curious, you know, when you had open ended textboxes, I would imagine that some students were putting in things like, you know, Darth Vader, or the Harry Potter or Paw Patrol. And I'm curious if things like that are gonna fall under fair use in an

AI environment? You know? Yeah. Yeah. How do you think about IP in that world?

the legal rights to it. I think there's a world in the future, where we allow kids to do that, once we have like partnerships in place. I think we're, that's not something we're like actively exploring right now. But I do think that it could be a meaningful way of like getting at our ultimate goal of motivating and engaging kids.

thought about it. And sounds like Fair Use doesn't apply in this case, at least as of right now. Yep. Fair enough. So let's zoom out a little bit here. I mean, we've already been obviously, we've been talking about AI in education, in a broad sense already. But it is such a hot topic right now, there was a really interesting article just this week with interview with Deborah quaza, who's one of the managing

partners of GSV. And a, you know, Ed Tech luminary and longtime investor, talking about how platforms built on AI and built on chat GPT are really going to disrupt teaching and learning and deliver better teaching outcomes, and they can engage, and I think that that's becoming sort of the prevalent feeling among especially the investor community, but I think, you know, some of the EdTech observers as well saying, Wow, this is a really new world of

edtech. There may be many, many startups and many, many ideas to come to really help education and AI, you know, play nice together and improve student outcomes as well as engagement. I'm curious, you know, again, you've been ahead of the curve on this. You've been doing this. Since before it was on everybody's lips. What do you envision? This new world might look like? You know, if we have this conversation a year from today, what do you think will have changed in AI and

education? What will exist?

If there's actual fear around cheating, for example, totally valid, that makes actually a ton of sense. And we're seeing, we're seeing it happen actively. But the hope is also really valid. In my opinion. There's some folks who believe that a lot of the problems in education right now, and I'm one of them can be solved, or at least mitigated through the support of

AI and AI enabled tools. Here's the thing, I think that this technology, these technologies, are they plural, because they're many so called category like, our language learning models are one type of AI. But there are a bunch of other like advancements happening in AI every single day

and week. So one thing I believe is that AI will solve real problems for teachers, were going to see that a number of different domains on a content level like koala, it will become more personalized and more differentiated, we will see AI as a phenomenal tool for one to one for scaling, feedback. Cool. But for example, like it's been ahead of the curve in this regard, anyone can get grammatical edits on their text. And so the paradigm of is just becoming a lot more accessible.

And then yeah, the creative partnership of it is a particular use case that I'm fascinated by Ricardo, you get an AI to like be or copilot. But here's another meta one that I think is going to continue to scale is how do you teach students to use AI ethically and responsibly? There is there's a teacher who blanking on the name he taught at the college level and he did a series of classes

about how to write with AI. And it was awesome to see like he was encouraging the kids to like use chat up And then he talked them about like prompt construction, and then how to evaluate the quality of work that AI is giving you and how to like, get it closer to what you want. And that I think is like such a cool way of introducing this into an educational

context. And it doesn't actually have to do do with any specific edtech tool, it just has to do with the teacher embracing this technology and thinking creatively about how a student can use it. Ultimately, my hope is that there's a variety of tools that educators can choose, whatever it is that they want to use, there are some teachers who will fully embrace it, like the folks who will maybe love that

tech right now. And some teachers who will fully reject it need to focus here like teaching and Waldorf that are schools. And that's fantastic. I think the wonderful thing about that is that we shouldn't be living in a world with a diversity of learning environments. And hopefully this will mean enables more diversity and learning environments. Yes,

validating. Because this is obviously a moment when my take on the podcast, anybody who's been listening will know is I am trying not to feel the fear and trying not to let the fearful side win because cheating and plagiarism, I think, are very short term concerns compared to the potential of this, but at the same time, you're right, they are absolutely justified

and fair, and reasonable. And they mean a lot, you know, you don't want your students finishing your, you know, French history class, not having learned any French history, because they got chat GPT to write all their papers, for them, that is a fail as an educational experience. So I love the idea of sort of seeing

both sides of that. And then, of course, to your point about all the different AI tools, really exciting to think about these copilots these opportunities for individualization personalization, differentiation, and you know, I'm glad you're mentioning the other types of AI as well, because some of the edtech companies we've talked to, on this show, use a an AI in very

different ways. I'm thinking of like teacher effects that, you know, literally records the text of classes, and then uses AI to generate reports about how much the teacher was speaking how much each student was speaking recommendations for how to make that happen more, it literally can look at whether the questions were open ended or not. And that is such a different use of AI than you know, anything. We've been talking about what MIT journey and Cha GPT. Let me ask one more

question. And then, you know, this is obviously a topic that is endless. You've talked about the hope and the fear. What is your personal sort of biggest hope and biggest fear? What do you think is the most exciting promise of AI right now? And what is the thing that has you most concerned that you think we should truly keep an eye on no matter what else happens?

enabling that to happen. And what I'm worried about, there are two things I'm worried about. One is, I think there's a there can be a school of thought around AI being used to replace educators, not augment educators. And that worries me a little bit. One, I find it philosophically strange. And two, I don't think AI is ready for that just yet. And yet another philosophical piece, kids need support and guidance of love from an educator. And I hope we never think about this

as like replacing educators. The second thing that worries me is strong stances, either super pro or super against. Right, I mean, he we need more neutral stances, and centered on educators needs a failure mode to me, is just as an example, a district being overly prescriptive in one way or another, you have to use it or you absolutely can't use it. And I'm not saying that that's like going to be a big trend. But it does worry me that like, there may be cases like that coming up.

chaotic environment. And I think your concern makes a lot of sense as well, the idea that some people may see AI as Oh, we'll make a chatbot that will be the teacher will make an artificial agent that will do all the education and we don't need teachers to grade essays anymore. We don't need teachers to introduce subjects anymore. And that is a very slippery slope and problematic on a lot of levels. I'm sure most of our listeners would agree. That said there's a lot of potential of

course. So we wrap up you know every interview with two questions as you know, what is a trend you see rising in the Ed Tech line? Escape right now. And just for the sake of this, because we've talked AI this whole time, let's take AI off the table. What is another trend you see rising in the landscape right now?

educators. And then actually the educator needs versus like, you know, the traditional way in which products are built and sold. So I'm super excited about that center of gravity, shifting more and more towards teachers and students for our tech products.

of life. So highly recommend that