Edtech Insiders
Edtech Insiders
The Rise of AI Companions: What Parents and Educators Need to Know with Dr. Sonia Tiwari
Dr. Sonia Tiwari is a Children's Media Researcher focused on how characters shape learning experiences. With a background spanning animation, game design, and a PhD in Learning Design & Technology, she advises edtech and media companies on ethical, developmentally aligned character design. Her work has been featured at Stanford, Harvard, MIT, UNICEF, and the National Geographic Society.
💡 5 Things You’ll Learn in This Episode:
- How AI changes children’s relationships with characters.
- What makes character design effective across media and education.
- The risks and guardrails needed for AI companions.
- How to use AI to support—rather than replace—creativity.
- What ethical, developmentally aligned AI design looks like.
✨ Episode Highlights:
[00:00:00] Why children instinctively learn through stories and characters.
[00:02:03] How character design works in animation, games, and learning.
[00:07:14] The overlap and tension between entertainment and educational media.
[00:14:35] What happens when beloved characters become interactive AI.
[00:17:33] Turning existing IP into AI-native characters.
[00:21:00] Why interaction length and purpose matter for children's AI use.
[00:29:23] Why design is the fastest path to safer AI for kids.
[00:31:33] How AI tools can self-regulate for young learners.
[00:36:29] Whether entertainment characters can power better edtech.
[00:40:23] Using AI to enhance—not bypass—creative development.
[00:45:39] Teens creating AI characters and the potential risks.
[00:49:18] A framework for deciding when AI is actually needed.
[00:50:11] How industry–academia collaboration could raise quality in edtech.
😎 Stay updated with Edtech Insiders!
Follow us on our podcast, newsletter & LinkedIn here.
🎉 Presenting Sponsor/s:
Every year, K-12 districts and higher ed institutions spend over half a trillion dollars—but most sales teams miss the signals. Starbridge tracks early signs like board minutes, budget drafts, and strategic plans, then helps you turn them into personalized outreach—fast. Win the deal bef
This season of Edtech Insiders is brought to you by Cooley LLP. Cooley is the go-to law firm for education and edtech innovators, offering industry-informed counsel across the 'pre-K to gray' spectrum. With a multidisciplinary approach and a powerful edtech ecosystem, Cooley helps shape the future of education.
Innovation in preK to gray learning is powered by exceptional people. For over 15 years, EdTech companies of all sizes and stages have trusted HireEducation to find the talent that drives impact. When specific skills and experiences are mission-critical, HireEducation is a partner that delivers. Offering permanent, fractional, and executive recruitment, HireEducation knows the go-to-market talent you need. Learn more at HireEdu.com.
As a tech-first company, Tuck Advisors has developed a suite of proprietary tools to serve its clients better. Tuck was the first firm in the world to launch a custom GPT around M&A. If you haven’t already, try our proprietary M&A Analyzer, which assesses fit between your company and a specific buyer. To explore this free tool and the rest of our technology, visit tuckadvisors.com.
[00:00:00] Dr. Sonia Tiwari: I think it's very comforting to read stories like it's a very natural instinct for children and adults to understand or break down a complex subject in the form of a story, and then characters carry that experience forward. So, you know, even young kids, they have favorite toys, favorite books, and the characters become almost these social others, these friends.
In context of traditional media, it was very cute and adorable for a child to say, oh, this book character is my best friend. But in context of ai, it becomes concerning.
[00:00:42] Alex Sarlin: Welcome to EdTech Insiders, the top podcast covering the education technology industry from funding rounds to impact to AI developments across early childhood K 12 higher ed and work. Find it all
here at EdTech Insiders. Remember to subscribe to the pod, check out our newsletter, and also our event calendar.
And to go deeper, check out EdTech Insiders Plus where you can get premium content access to our WhatsApp channel, early access to events and back channel insights from Alex and Ben. Hope you enjoyed today's pod.
We have a really, really exciting episode of EdTech Insiders today. Dr. Sonia Tiwari is a children's media researcher exploring the role of characters as facilitators of children's learning experiences. Her interdisciplinary background includes an MFA in animation gaming industry experience as a character designer, a PhD in learning.
Design and technology and research consultation for children's media and ed tech industries. She's been invited to share her research at the Stanford Graduate School of Education, the Harvard Graduate School of Education, the MIT AI and Education Summit, UNICEF Communities of Practice, and the National Geographic Society.
Dr. Sonia Tiwari, welcome to EdTech Insiders.
[00:02:03] Dr. Sonia Tiwari: Happy to be here.
[00:02:04] Alex Sarlin: So first off, your research is about characters as facilitators of children's learning experience and and characters have been central. Basically the core of children's educational media for many, many decades. Sesame Street, Dora the Explorer, bluey, SpongeBob, a A, anything you can imagine, but that's educational and entertainment media.
In your view, what makes characters such powerful facilitators for education?
[00:02:33] Dr. Sonia Tiwari: I think it's very comforting to read stories like it's a very natural instinct for children and adults to understand or break down a complex subject in the form of a story, and then characters carry that experience forward.
So, you know, even young kids, they have favorite toys, favorite books, and the characters become almost these social others, these friends. In context of traditional media, it was very cute and adorable for a child to say, oh, this book character is my best friend. But in context of ai, it becomes concerning because there's like potential of overuse addiction, and children can be obsessed with a book character and have their parents read it over and over again.
But that's because it's in combination. With a human, it has a limited scope, a limited end, like, you know, the book definitely ends at some point. It's not nearly as addictive. So yeah, we, we are just naturally drawn to stories and characters. It's just like characters on steroids when it comes to ai.
[00:03:44] Alex Sarlin: Yeah, I mean if you think of one of the big innovations of large language models is that it's AI that can use language in human-like ways, right?
And it's just to put it simply, it means that you can have an AI that simulates any kind of human, any type of character, human or non-human. And it. Introduces this world like as you're saying, a world of incredibly exciting character-based interactions, but also risk because some of the ways we've engaged with characters in the past have been in the context of a story or in the context of another human facilitating.
The idea of being able to sit and talk to your favorite characters endlessly is this really strange double-edged sword. You know, you've studied character in a lot of different contexts. Tell us a little bit about some of the design choices that go into characters, because obviously people spend a lot of time, a lot of money making characters that are gonna resonate with children.
What does that look like? How are characters designed to connect to and help educate children?
[00:04:44] Dr. Sonia Tiwari: Right. So I'll, I'll give like the, the gaming industry version of it. We usually have a very detailed design brief, which is based on the overall experience of the game in films and tv. It's usually part of the show Bible or, or like a very detailed document that says, oh, this is the world.
These are the characters, this is their personality. This is our visual style. This is how we'll animate it. And so within all these contexts and constraints, the character designer then starts with like a bunch of different sketches trying to assess like what's the best way to express the personality of this character based on the brief.
And then once a design really speaks to the our director, the director of the story, then we develop like different views of the character, like a front side perspective, different poses. So we create like a character sheet and then we bring in like different voice actors. Usually directors have a tentative.
Understanding of understanding what would be a good voice for this character. And it's not like, you know, a big, strong character with a very muscular voice or tiny character with a tiny voice. Sometimes it's like the Trolls movie has this warm character, which is very small, but has a very huge voice and it's kind of funny that way.
So yeah, it's like a creative decision to think of the voice, the clothing, accessories, personality. That weaves all these environmental and backstory details and visual and audio forms, and even the style of the animation.
[00:06:24] Alex Sarlin: Uh, I always get very excited by some of the elements that the education industry can borrow from the entertainment industry.
And, you know, you mentioned the Trolls movie, which is an entertainment franchise, but we can talk about, you know, characters in the context of entertainment or education. I, we've seen Miss Rachel, who is a real person. Basically transform into a character on YouTube and now transform into a character that's being sold as in toy format.
You know, there's all this interesting interactions between educational media and entertainment media. You see things like Sesame Street that are crossovers between them that's educational, but you know, popular in, in, in a way that entertainment usually you only usually see with entertainment. Tell us a little bit about the relationship.
'cause you study this carefully, you know, what is the relationship between the educational media. World. World and the entertainment media world when it comes to character design.
[00:07:14] Dr. Sonia Tiwari: I think like foundationally, I have this problem with like edutainment and like entertainment education. Like the equivalent of that in writing, for example, is let's say like a good understanding of grammar, whether you're writing a novella or the script of a documentary.
Or anything else, like there are some foundational skills that cross genres that are agnostic to any ways in which it'll be used.
[00:07:44] Alex Sarlin: Mm-hmm.
[00:07:45] Dr. Sonia Tiwari: So I think like foundationally character design is the same in entertainment or education or anywhere else. It's the application that's different. And so that's why like I, in my one sentence, which I say character as facilitators of learning experiences, they could be purposed as just facilitators of entertainment or information seeking or anything else.
So yeah, like the foundational understanding of characters as a visual or audio or a mix of different media representation of a personality. Something that weaves in background, something that helps children. Visualize a concept, for example, like the characters in the Inside Out movie Yep. Is just like a very clear correlation between the emotion and the character, but sometimes it can be indirect.
There's this children's book I love called, where Oliver Fits, and it's a puzzle piece called Oliver, which just goes through life, you know, finding where do I fit, and he finds different puzzle pieces, never gets along anyone. In the end, like finds a whole place full of misfits and finally fits there. And so even like, you know, it could visually represent this idea of not fitting in and it can serve as a powerful metaphor.
And so yeah, like in, in those ways, whether it's an educational story about belonging. Or understanding of emotions or just purely entertainment. Like the minions, they're just like silly, goofy pill like characters, right? Who are just like exaggerated and, and kids don't like separate media. Like this is the adult trying to distinguish between education and entertainment.
Kids just perceive, uh, you know, like Sesame Street, it's short, it's educational and it's not chocolate covered broccoli or whatever other descriptions we have. It's just like play and exploration for kids, whether it's media or toys, it's composite effect. It's not separated for them.
[00:09:52] Alex Sarlin: I really like that distinction of characters as facilitators of different types of experiences because you could have the same character facilitating an entertainment experience, facilitating an educational experience, facilitating even a, you know, a social experience or a therapy experience, you know, in some ways.
Or, I mean, you can, the sky's the limit because. Character is a representation of a person and people can fit in many different situations as well. And I also, I love what you're talking about with this idea of, you know, the puzzle piece character or, you know, I grew up, you know, I have a background in in Sesame, so I always go back to Sesame as sort of educational characters and a lot of those characters were designed to sort of represent different aspects of childhood, right.
The greed, the sort of insatiable greed for candy and cookies that is Cookie Monster or the insecurity of Big Bird who's like enormous, but is actually a very young child at heart and attached to his teddy bear. Like those characters, like the puzzle piece and, and like a lot of the ones you're mentioning are designed to physically embody particular aspects.
Of the world that they want to communicate to the children. And that's true in entertainment and in in education. But you know,
[00:10:59] Dr. Sonia Tiwari: and there's also the, I love the idea of abstract design, right? Like I, I did a research fellowship at the Strong Museum of Play and they had a whole exhibit about the history of sesame and how each character design came to be.
Yeah. And I really love that. Because Sesame was trying to appeal to a very like wide range of audiences, like the red color of Elmo. Anyone can see themselves
[00:11:25] Alex Sarlin: exactly in
[00:11:25] Dr. Sonia Tiwari: Elmo or any other character because it's just, you know, kind of like a frog esque, random, abstract
[00:11:32] Alex Sarlin: character. Totally.
[00:11:33] Dr. Sonia Tiwari: And even like I was listening to this documentary about Hello Kitty and the character designer there, she was talking about.
How, hello? Kadi does not have a mouth shape because she wanted kids to imagine the emotion themselves. And that is so powerful. Like I, I think like sometimes we get too prescriptive that this is a character, that's what it's feeling, and that's the magic of toys and that's why I'm also concerned about AI toys.
Now that leaves something to the imagination, like kids pick up dolls and have dialogues between them, and it's a way of practicing communication. When the doll is doing all the talking and moving, and the kids are what? Just like sitting and listening, that takes away the imagination. So it also depends on what the character is, what the purpose is, what the context is.
[00:12:23] Alex Sarlin: Yeah, it's a great point. And that, that sort of strategic withdrawal of particular aspects, like you're saying hello, kti doesn't have a mouth so people can figure, can put themselves into it. Elmo is specifically not, you know, and all the monsters on Sesame Street don't represent any particular race, but they are gendered, there are characters who are non-gendered on purpose.
So people from, from either, uh, either. You know, any, any gender background can relate to them. I, I remember reading about Spider-Man and Spider-Man is sort of a famous example of a character who, while technically Spider-Man is a white male teenager, because he's completely covered and he's completely covered in red.
People projected themselves into it. And a lot of young black comic book readers always sort of were like, I, I think Spider-Man's really, really black, or really, really Latino. Right? Because, because he's, it's a character that went in costume, sort of has no signifiers. It's a really powerful point. So.
We've mentioned AI already a few times, but we should really get into it 'cause this is, I think, such a crazy moment for character in the world. I mean, you mentioned that characters are usually designed in very specific contexts, right? If you're making Toy story, you design a little, you know, pig, you know the dog Slinky character, and you give it a voice and it's Jim Vardy and it looks like this and it moves like this and it looks like this from the side.
But you have total control over it. You know exactly what scenes it's gonna be in. You know exactly what words it's gonna say. You write the dialogue. That's how traditionally how characters have been designed. They're designed and then put into a story or a context. We are now in a world where you can basically design interactive characters that can respond and actively reflect what somebody is dealing with.
They can converse, they can be led by a student or by, by a kid, or by, by anyone. You have sites like Character, AI and replica, and lots of other ones that allow characters to be created from whole cloth in no context, right? Without a story necessarily behind them. This is a really strange moment. Can you just before we get into the details about, you know, children's learning, like what does it feel like for somebody who understands character in the way that you understand it, to see this sort of potential transition happening of sort of character without story?
It's really strange to me.
[00:14:35] Dr. Sonia Tiwari: Right. So it is like the, the way we are usually groomed and conditioned to experience characters is like you said, in that context. And on another podcast I was asked that you know, which one of your favorite childhood characters. Could you see possibly being turned into an AI character?
And would you actually enjoy it? And I was thinking back to Powerpuff Girls. Yeah. My favorite character was Bubbles. Yep. And the thing is that the entertaining part of that show was the dynamic. Like one of them is a kind mature leader. One of them is like a, you know, like a browdy. Character and then you know, bubbles.
So it's the comic relief of the dynamic. And so on its own, it would be really annoying because there's no opposing. Element to counter it. If you think of like, you know, monsters Inc. It's like Mike Wazowski in contrast to the big Sully, that that is the appeal of the character or like Mandalorian with Roku.
That's the appeal. Like either of them on their own or not as amazing. They're amazing in contrast with each other,
[00:15:45] Alex Sarlin: right? I agree with you, but I think that presumes that the relationship between a human and an AI character is always one-to-one, and that, that there's no rule about that at all. I mean, you could have a conversation with all three powerpuff girls and anything you ask them, they bounce off of each other and all react as themselves.
There's, there's absolutely nothing keeping us from doing that.
[00:16:04] Dr. Sonia Tiwari: Yeah, I mean that's happening too, but what I'm saying that at least as of now, there's a lot of one-on-one in isolation. There's this company called like Native Voice, which is Yeah, basically, yeah. Right. Like franchising existing characters.
They do train the chatbots on like extensive documentation provided by the creators, so it's, doesn't say anything outside of the story world. It doesn't say anything that is or, or like repeats phrases that have been in other formats. Film, TV games. Right. Even within those confines, because we are conditioned to experience characters differently, we're also in this transitional phase.
[00:16:45] Alex Sarlin: Very much so. So
[00:16:46] Dr. Sonia Tiwari: even if we see like these three characters together, it is for people like us who have grown with other types of characters, it'll take some. To transition into this new reality. Whereas, like, you know, buddy, I was like earlier on the podcast, like our promoters of like native AI content.
Exactly. Although that's not, again, like, that's not the only solution because all media franchise, like, imagine like, uh, Disney World for example, they can very easily launch like in person booth or hubs where you go in and chat to these characters. It's, it's gonna happen whether. We like or not like, we can't say that let's make all AI characters native to ai.
There is going to be re repurposing and recycling as well as no
[00:17:33] Alex Sarlin: question. And you know, so just, just to add context for people who haven't heard any of this, you know, we interviewed the founder, Ivanov of a company called buddy.ai. They have a native character, a character who is designed to be an AI character designed basically in pieces to be conversed with and talked to by a student and doesn't really have.
An existence outside of the AI doesn't have necessarily a backstory. It doesn't really appear that much in non-interactive context. And I have a theory that AI native characters that will be created after the advent of AI for AI will have a very different look and feel than ones that are created before AI and then translated into it like this Native Voice or like, you know, you can imagine any existing property.
Potentially being turned into a interactive ai. But I mean, just a question, you know, the context you'd name about, like, okay, let's say you're doing Neo from the Matrix and you're talking to him and everything you ask him, he's gonna respond in the context of the Matrix movies that already exist, or he's gonna quote something from the Matrix video game.
Like that actually doesn't sound like a particularly. Fulfilling experience compared to much else. So I agree that they'll exist. But don't you think that the limitations that the IP limitations that exist on existing characters are gonna constrain the possibilities of ai?
[00:18:51] Dr. Sonia Tiwari: I mean, I don't think so because if we look at the history of how other characters expanded their transmedia suite.
Barbie started as a toy, but then moved into TV and moved into other things. Now I don't know how the AI is beep, and there have been other crossovers, like something started as a toy. Something started as a TV first, then became a popular toy. Transformers, uh, is another example of toy too, like this, this movie franchise, right?
So we have seen these kind of carrying over the experience of a character across platforms. So it is going to, like, as of now, of course, it feels a little creepy to experience characters that we are so used to seeing in other media. It, it also happens with. Purists who love a book, when they see a movie based on it, they're like, no, no, no.
This is like nothing like the, like the book or like dyslexics like me. When I see a movie first and then read the book, I'm like, what doesn't like, I prefer the movie. So it just depends. So with ai, it's probably gonna be similar. There are going to be people who find it more interesting to interact with a character and discover a story.
Through that interaction, organic interaction, very non-linear. Whereas there are going to be people who prefer the linear model. Like I, I totally personally hate it when people launch a whole series and then there's a prequel, then there's a side movie, and then there's like, I, I like chronological just.
Deep sense of things, whereas some people prefer this kind of organic open world game modality where, you know, just enter, explore, figure out.
[00:20:36] Alex Sarlin: And I think once that happens, the floodgates are really open. I think whoever makes that work to take commercial pre AI native characters and use them in an AI native context, and they figure out how to do it safely and in a way that works for customers, I think that's gonna create a whole new genre.
And there's gonna be a whole lot of other people suddenly doing that as well. It'll be none of it, and then all of it all at once.
[00:21:00] Dr. Sonia Tiwari: So I, I'll start with the acknowledgement that neither me nor any one researcher or thought leader will have all the answers.
[00:21:08] Alex Sarlin: Sure.
[00:21:08] Dr. Sonia Tiwari: I think like the best part of research is being curious and trying to find, based on the information available now.
That said, like there are some like common sense things. For example, even the quality and length of interaction matters. So giving some kind of like, you know, blanket recommendations that no anthropomorphizing or don't introduce AI characters to young kids, or let's not use AI at all. Like those kind of, uh, simplistic solution to a systemic issue usually never works.
Got it. So it's more like, okay, even if it's a open-ended character interaction, what are the guardrails around it? Is there something built in the product? Like is there an average interaction Time? What I like about Buddy is that, or, or other like tutor type educational AI characters is that. The goal of learning is set upfront, so there are no surprises.
You, you click on a color or shapes module and you know that that's what the interaction is gonna be about. Whereas sometimes in the open-ended chat bot, it's like, even if it's, there's some general guardrail about, I'll not talk about anything inappropriate. Even within that, the conversation can go anywhere and it can last however much you want it to last, and that's what we saw in.
Social robot type of toys, like Moxie. There were some amazing use cases like, you know, in hospital settings and in therapy sessions in the presence of a mental health expert. But again, at home, if no one is watching and a child is spending hours talking to this supposedly educational toy. Just the amount of interaction makes a difference.
I know like in children's media, there's this debate about, oh, it doesn't matter, like screen time is not a good metric because it depends on what you watch. Sure. But also the time. Right. Like you could be watching nature documentaries for eight hours, so it's informational, but also eight hours is problematic.
Right? So it's kind of like that balance, uh, of having a holistic conversation about what is the character design in service off in the overall learning experie.
[00:23:26] Alex Sarlin: Yeah, it's so interesting to hear you talk about it. 'cause as you were talking about the sort of blanket roles, you're like, well, you know, maybe the idea of no character is under a certain age.
And I, I found myself nodding along. I was like, well, it actually feels like some sort of simple heuristics. Like that could be instructive. They could push the field forward if it's like, you know, but you're right, it also becomes very constraining. It becomes very blanket and nuance is, is necessary at the same time.
I mean, we're in an era where we're seeing schools, you know, ban cell phones outright. In dozens of states, and I feel like they tried these nuanced approaches for for a decade of like, oh, you can have it during this time and not that time, and you can have it in at class, but have to keep it in your bag.
And finally, they've sort of reverted to this extremely. Yeah, a simple rule, and I can't help but wonder if, even if it's not the actual right move to say, you know, no more than two hours a day, that doesn't really make sense. At the same time, it's a rule that everybody can understand. Right. Do you feel like there's a sort of balance between that kind of sweeping, sweeping simple heuristic versus the actually trying to get to the nuance and get it exactly right.
[00:24:31] Dr. Sonia Tiwari: Yeah, so I recently published an article in the AI Brain and Child Journal that kind of looked at different factors that we can consider in design. So just, you know, starting with the child's context, like what is their developmental profile? What is their capacity to understand in, in context of a AI character driven learning experience, and what are their needs?
Do they have any learning challenges? What are their support systems at home, et cetera. Then we kind of look at the character design factors that given this profile of a child, what kind of design might appeal to them, might be developmentally appropriate. What is the visual and audio details? Are we representing a culture and issue, a concept idea, that sort of thing.
And then once we have the character and child's understanding, then we can look at the interaction factors. And so even just the scale of interaction, it could be a simple information retrieval like an Alexa type smart speaker that, you know, what's the weather like? Set an alarm for such and such time.
It can be very transactional or it could be a friendly mentor like Buddy or LO or any of these like tutor type characters where there is some personalization and there is something more than just information dumping, but it's still very much structured and within a very clearly specified goal. And on the other extreme is what we're seeing, where the problems are, where they're extreme parasocial relationship and companionship.
And that's one of the reasons why I say I use the term AI characters as opposed to AI companions, I think, which is more popular because my belief as a character designer is that not all characters are designed for companionship. Characters can yeah. Serve different purposes. Uh, in Conmigo, for example, the character piece is just, uh, icon.
Who winks occasionally, like, or they're very like three or four different states. Uh, not even any detailed animation or a body, it's just a chat icon with eyes. And it's not designed for companionship. So that's how we can start thinking of, uh, characters as the bigger umbrella. And underneath it are all these different applications.
[00:26:49] Alex Sarlin: That's a really interesting distinction when you talk about that sort of framework of, of understanding all these nuanced elements of a child's developmental process, what they can handle, what if they have learning differences, sort of all of the individual differences as an arbiter of how they can use characters.
You know, my first instinct is. That's a wonderful framework, but how would you implement that in the real world? How are you asking parents to develop that? Are you asking schools to have to evaluate that? But then I, I realized that we're in this interesting AI era, and you could actually have that kind of framework built into the AI tool itself, right?
If you had a. An AI character, I'm not saying companion, right, an AI character. It could actually, in response to how it's being talked to in response to what it's learning about the student user start to regulate itself or change what it does or sort of fit the conversation into that different types of patterns you're mentioning.
Is this information seeking? Is this, you know, emotional is this, and that becomes very intriguing because it's the kind of complicated context that is very hard to implement from the outside. But from the inside actually, maybe could be quite manageable.
[00:27:57] Dr. Sonia Tiwari: Right. So I, I wrote the article for designers actually, and as a consultant, I work directly with the creators because my belief is that reform at scale can only happen through design because we are trying to control a global technology locally.
Policies and reforms and, which is amazing. Like I'm, I'm not like throwing shade at the efforts of policymakers, but it's a very slow process depending on where you are and what kind of support you have. Europe, for example, in general, is more interested in creating rules and regulation faster. At the same time, like in Silicon Valley here, as you know, the culture is like move fast and break things type of, uh, attitude.
And we often see that technology is moving much faster than laws and regulation can keep up with. So at this time, the only way to make impact at scale is to reach out to the designers so they can build all these guardrails right from bottom up within the product itself instead of, you know, after the fact.
Creating foundations and frameworks and lawsuits. Totally. And putting out fires later on. But then again, because it's a request, right? The kind of people who reach out to me for consultation are already like, you know, former educators, turn CEOs who want to do the right thing,
[00:29:23] Alex Sarlin: right?
[00:29:23] Dr. Sonia Tiwari: Whereas a lot of, you know, these tech bros are MBAs with no background in education and they approach it like any other product like.
Education is just a domain. Like they don't think that subject matter expertise means anything. They've treated it like any other product, which is challenging because children are a vulnerable population In any other product, if you release a half-assed thing, the consumers will revolt, they'll write negative reviews, and children don't express themselves like that.
That's
[00:29:53] Alex Sarlin: true.
[00:29:54] Dr. Sonia Tiwari: So it's a vulnerable population and we, we've just like unleashed this technology on them. And so I try to think of guardrails at this three levels. The biggest one, of course is design because there's like, like you said, you know, there's not much we can do after the fact. Yeah. And then the other two that is possible without design is just building a culture around how we use ai, how we interact with AI characters, for example, with young children.
We can create simple, this culture that anytime we interact with a new AI toy. There's gonna be a caregiver or educator around us to do it together and monitor in real time. And if there's a problematic conversation, just unplug it or do something else. And then, you know, at the legal level, again, that's something that can be done outside of design.
Because when it's just a request, like I said, only the ethical pro ethics founders are going to listen for everyone else. You need a firm guideline that this has to happen or your product will not ship. That's the language they'll understand.
[00:30:59] Alex Sarlin: Right. And, and it strikes me that even like the, the role you are mentioning about, you know, maybe if a child is under a certain age, they need to have a parent present.
That is true in various types of consumer products for children. But with an ai, you could actually bake that into the system itself. You could have the AI say, oh, nice to meet you. I'm, I'm ex character, and the kid is four years old. And it says, oh, where, where's, you know, I'd love to talk to your, your parent or caregiver.
And if they don't hear a parent or caregiver, they'd say, oh, sorry, we can't talk right now. Come back when you have a parent or caregiver. Like you can literally break these things into the system, you know?
[00:31:33] Dr. Sonia Tiwari: Right. Yeah. And so that's one of the things in the engagement that the interaction branch in my framework is like designing for joint media engagement.
But again, like, because very few. Founders are willing to adapt these practices, like Yeah. Then after the fact we'll have to build the culture like this on our own. Yeah,
[00:31:52] Alex Sarlin: exactly. No, it's really interesting. So you know, you mentioned like LO and Buddy or the Conmigo type of sort of pseudo character and Conmigo.
We obviously have Duo the Owl, but you know, it always strikes me as somebody who also loves the idea of taking certain elements of the education world, certain elements and, and using them in an educational context. It's always sort of bugged me that there are so few. Legendary educational characters, right?
I mean, we can talk about the Elmos or Big Birds or Doras. There's a handful of educational characters that are sort of transcended, become like household names, but there's thousands of characters from the entertainment world who are household names. In fact, it's, it's a huge part of kids. World are these entertainment characters?
I'm curious, and we can say this in the context of some of the things we're talking about from the AI ecosystem. I'm not trying to totally change the subject, but I'm just like, what would it take to have an educational character or a character coming from entertainment into education that could become sort of a ubiquitous character, a household name character that can almost symbolize the space?
[00:32:56] Dr. Sonia Tiwari: Yeah, so I think we need to take a step back to see the bird's eye view of how things work in the entertainment industry. It's usually, it's a business, right? It's a for-profit business. So usually the quality of designers that are hired by game studios, it, it's like a very challenging portfolio review process.
Extremely rigorous process through which these designers are selected, and then they even hire different designers. To align with different aesthetics. So my style was, you know, very 2D Kauai type of drawing style, which works well with mobile games because the screen size is small and you need like icon, like very clearly understandable characters and small scale.
Whereas like I have friends who are really good at anatomical drawing and they design more complex characters, so they are hired for more of the 3D realistic type of simulation games. So that level of rigor of design is usually not present in the. Education unless it's, you know, a place like pbs, which receives millions in funding, used to receive millions in funding for National Science Foundation or, you know, c PB type of places.
So it, it's also like where the funding is coming from, how much funding is coming in. And so PBS is one of those rare places where they do partner with proper animation studios. So they do have the artistic rigor. And they also partner with all these curriculum advisors and education experts. So there's the educational rigor combined, whereas in most cases, like you know, many of these well-known shows on YouTube, their metric is different.
They are not grant funded and they don't care about social impact. It's more like. We need 2 million more followers this month. What will get more eyeballs? Let's increase our saturation and decrease our script quality. Like whatever it takes to get there. So yeah, it's the broader, unfortunate societal context that where quality programming is coming from.
[00:35:07] Alex Sarlin: I mean, we do a lot of Disney plus these days 'cause I have very young kids and it is just astounding how many shows on there. Just the production value, the rigor, the thoughtfulness, and I don't mean educational thoughtfulness of course, but the thoughtfulness in which they're designing these ecosystems to just be like.
Absolutely. I mean, irresistible to their core audience is just incredible. And I, I think everything you're saying really resonates. And so let me extend a, a question to you. I mean, you know, assuming we do get some of these frameworks you're mentioning, right, some of the nuanced frameworks about information seeking, about time limits, about age limits, about really baking these into the design and into the post-launch sort of nuances of the laws.
Do you anticipate that there will be more and more educational characters that are sort of exciting for kids? Or do you think it's more likely, I think I know where you, where you'd say, but let's see, that you'd have some of the entertainment world characters sort of migrate into an educational context.
You can imagine a kid having a sort of an AI tutor or an AI support bot, learning companion, whatever you wanna call it, and being able to skin it as a Pokemon or skin it as a, a sesame character, skin it as a, as a care bear, whatever you'd like. Of course what's happening under the hood in that case is still the educational tutor bot, but you can use some of the nuances of the character.
Do you think that's a potential future or are these worlds sort of too hard to combine?
[00:36:29] Dr. Sonia Tiwari: No, it's definitely possible and like I said, like we have seen these type of crossovers in the past where popular movie or you know, like there are like educational children's books and TV shows with Hello Kitty for example, and start out as an educational character.
So that has happened in the past, and again, like I'll zoom out to that economic version of it. It depends on like who's invested and what the aims of the organizations are. And even like, I'm not even saying that, oh, if it comes from a, a grant funded, educator driven. Project, then it's automatically good.
Like of course there could be quality issues there as well, but in most cases, again, like it, it depends on the wider goals of the organization. Like Fisher Price, for example, is a nice example of how they've taken some of their toys, turned them into mobile app experiences that are educational and high quality because many of these companies do.
Hire subject matter experts when they're translating a franchise from purely entertainment or toy based into something else. So it, it just depends on like, who did they hire? What was their goal? Did they implement it with integrity? And even if, and like many times this process is not transparent.
There's like a marketing edition that is revealed to the world that, you know, 80% parents say they noticed a difference in their child's math skills. They don't say 80% of what, and like, how did you find that out? Was there a pre-test or study or is this a sample size representative of the population you're talking about?
So none of those details are revealed. So it's kind of like, you know, the back of a cereal box. The big font says protein fortified, and the small font says 200 grams of sugar. Depends on, you know, what the real ingredients are.
[00:38:26] Alex Sarlin: It's a great point and I think, I think it's a good answer. It's like it's possible there are contexts in which it has happened, that sort of entertainment, education, crossover, but you have to sort of follow the incentive structure all the way down to see, you know, where it will happen and where it will make sense and.
What kind of characters people would say. This is a great additional channel and it fits with our mission and vision as a company, and which ones would say we protect our characters IP at all costs and it only costs a huge amount to appear anywhere, and we would never license it to this context. Like there's a lot of economic.
Factors, as you're saying, that would go into any of these decisions, but I am holding out. I think it's exciting. I think it's exciting, and I think there may be ways, again, to combine the sort of intelligent, nuanced, pedagogically sound. You know, brains inside some of these LLMs and ais with the looks and voices and music and settings.
Of entertainment. Uh, I, I, I feel like that's a really
[00:39:21] Dr. Sonia Tiwari: exciting AI quest. The AI literacy game is a great example of this collaboration between Stanford and, and Google.
[00:39:28] Alex Sarlin: Yeah.
[00:39:28] Dr. Sonia Tiwari: Turned out great and like it, it's one of the good examples of when professional game design skills combined with solid curriculum.
[00:39:37] Alex Sarlin: Agreed AI literacy curriculum. Everybody who hasn't seen that, that just launched a, a couple months ago, it looks and feels like a professional game. Uh, it really does. It doesn't feel like it's a sort of subpar made for the educational market experience. It feels really, really, uh, powerful. So, you know, let's talk about creativity.
You, you, you've written about this, you've written about children. As you know, playful artists and children have all of this creative energy and juice, and these AI tools at least have the potential to allow children to express their creativity in amazing ways. Especially regarding characters. What do you see as the relationship between kids' creativity and ai tools and characters?
How might they come together in the next five or 10 years to create hopefully some really interesting and and positive experiments and experiences?
[00:40:23] Dr. Sonia Tiwari: So one example is we've seen AI characters kind of facilitating some kind of creative activity, even like prompting children that let's make a story together.
Like most AI toys have that feature of like, you know, the AI would say one sentence and the child adds another and build off of each other. Or, I think like the straight up prompting art does not count as being creative in my opinion. They're just, you know, sitting like. Write a story for me or create an image of a dog made of a moon, and then people applaud that, oh, the child is thinking about a dog made of moon.
And that's so creative. Sure. Like, it, it, it has some origins of a creative process, but creativity is play, it's exploration. It's, you know, expressing your raw ideas in different ways until it evolves into something that pleases you and for that organic process to take place. There has to be something more than just prompting.
So I'm not like anti AI in terms of creativity. I just feel like there's a foundational skill building process that shouldn't be bypassed before we move on to AI for augmentation. So I've heard children say that, well, I'm not that good at drawing, so I'm gonna use AI for that. And then I say, well have you, like, even if it's just stick figures, have you tried drawing first?
And there are very like short courses you can watch on YouTube to improve your drawing skills. And you may not be professional, but that's not important. Like expression is not perfection. So it doesn't have to be a fully rendered pixel perfect image to get an idea across. It could be stick figures. And so just, you know, exploring that foundational creativity first and then seeing where.
AI might be helpful, and one of those examples was where. I was working with this school whose drama club budget was gone because of, you know, all the art programs being cut off and so they still wanted to act out the play. They just didn't have a costume budget left. So what we did was we recorded them with cardboard props, so they still gotta act, perform, and then used PIKA video to video to transform their cardboard costumes into more fancy ones.
And then had a film festival.
[00:42:43] Alex Sarlin: Amazing.
[00:42:43] Dr. Sonia Tiwari: And so something like that, you still get to rehearse and embody these characters, act things out, the whole experience of your school coming together. So the human experience is not cut off, but AI was used more thoughtfully.
[00:42:58] Alex Sarlin: That's a really fascinating example, a really, really compelling one.
And you know, I, I think you're putting a point on something that is, I think, gonna be increasingly important for the creative arts, which is it. It's just a very strange moment that, you know, I think we're all struggling to get our heads around, and people feel this in creative writing. They feel it in drawing and art.
They feel it in music, they feel it in filmmaking. You know, EU suddenly have these tools that can turn ideas into professional quality work. That's just never existed before. And I think on one hand people get excited about it. I, I certainly get excited about it. I say, Hey, that kid who said, oh, I can't draw.
I'm so sad I can't draw, but I have this amazing idea for a comic, but I can't draw, so I'll never do it. Suddenly, hey, there's nothing stopping them. That's really exciting to me. But you're right. At the same time, some of the. Creative work that, the work that comes into actually building your skills, the collaboration that you mentioned every, the school coming together to do the play, you know is, would be completely lost if you have a single kid or two kids sitting there saying, we're gonna make a play and we're gonna make up all the characters and we're gonna invent actors.
Like there's a really interesting push and pull between the different benefits of creativity that I find it extremely exciting actually. It's like exciting to navigate, but also. As with many things, there are ways for it to go wrong. You, you can imagine AI for creativity that ends up squashing collaboration.
It ends up squashing squeal, building it. It turns everybody into just a prompting. It's like, who can create the cleverest prompt or the cleverest reference in their prompt and then they decide they're the best artist? That's a very strange world. I
[00:44:28] Dr. Sonia Tiwari: think it can be used thoughtfully as a small part of the brainstorming process.
So seek inspiration from nature, from personal life, and maybe a few AI generated outcomes, and then do a mashup of all these different ideas, because a lot of creativity is remixing, like there's hardly ever something completely original. So instead of just like, you know, stopping there, like, oh, a kid who can draw used AI to generate an image.
How about then building on top of it and asking AI to even like break down the process and then trying to create something similar, even if it's not the same on your own? Like does that inspire you to then learn or that's the end of it that, well, I can draw, therefore I'm always going to generate, or this inspires me to express myself in other ways.
Maybe drawing is not my thing. Maybe I'll write a poem. Well, I'll do something else.
[00:45:23] Alex Sarlin: So just to put these streams together, you know, what, what do you think it's gonna look like when students actually start to create their own customized characters in an educational context? Right? Uh, students are as capable as anyone of creating characters.
I'm curious what you think about that.
[00:45:39] Dr. Sonia Tiwari: Yeah. Students obviously have the capacity to imagine stories build their own characters, especially teens who have access to building custom gpt and other such characters. Again, there's the challenge of, like we saw on character ai, it is even possible for kids to go there and take up all of their insecurities and desires and represent them through a character, and then obsessively engage with it.
So it's, it's kind of personifying your own mental health issues that are unresolved and unsupported and just reveling in it instead of stepping out and asking for help. So just again, like depends on the context. They are very much capable of building a tutor for their specific learning needs. A good example.
They're also capable of building like a, a fictional character that is again, used to facilitate learning or a psychological manipulator accidentally even.
[00:46:41] Alex Sarlin: Yeah. Yeah. I mean, I feel like such a through line of this conversation is that there's this exciting potential and then there are these pitfalls and there's these concerns.
Are real and that are nascent and that people are just beginning to get their head around. You mentioned you don't like the word AI companions. You prefer AI characteristic companions, and AI companions as a term has already become laden with all of this sort of baggage. Just I know.
[00:47:08] Dr. Sonia Tiwari: I mean, most of the things out there are AI companions, right?
Like that's why that term is more prevalent because most of these character, AI and replica type of chatbots are designed for companionship. So. Like that term is valid too. I think character to me is the bigger umbrella.
[00:47:26] Alex Sarlin: Yeah, that's fair. Well, I mean, I just, I think my last question, I know we're a little over time here, but my last question for you, just as EdTech companies start to navigate this moment of AI continuing to grow.
Usage continuing to grow, but also concerns continuing to grow, and people starting to truly be nervous about the ramifications of some of these technologies, including, especially in this moment, AI Companions. How do you recommend they sort of think they, they don't get too weighed down by the worries and also not too naively inspired by the potential.
Like how do you keep a balanced mind about what's exciting about this moment in terms of AI character and what we shouldn't run with until we know more? I
[00:48:09] Dr. Sonia Tiwari: think in like as a consultant, I do kind of like these goal setting exercises that you know, anyone designing an EdTech product can do for themselves.
Just list out what the learning goals are, what the learner context is, and then really think about which of these things, what's the simplest thing that will help us achieve this goal If it's just a simple paper and pen activity. Sure. Like even add a downloadable worksheet and that could be enough.
Some things might require a more gamified experience. Some things might require something else. So AI is just one of those things of all the functions that your product is performing, which function requires AI augmentation and which does not. And what happens is we have like this AI feature, diarrhea.
Everything there is will be layered with ai, even if it's not needed. So that's, that's the kind of assessment we can do. Even like creating a rubric for yourself that does this really need ai? If not, then you know, let's find something simpler. So yeah, just finding a more value aligned application.
[00:49:18] Alex Sarlin: That's interesting.
So it's like it start from the point of need rather than starting from the point of AI is here, what can it do? Let's do something fancy with it. Say, you know, stick with what are we trying to accomplish here? What are our educational goals? And in what cases might AI be valuable? But let's not always assume that that's the go-to answer.
So your dream vision of what might happen if education really did solve the underlying problems, right? If, if we solve some of the issues of addiction, solve some of the issues of cognitive offloading, solve the issues of, you know, who has access to ai, tutors and characters, at what age, how do we limit the usage?
How do we, you know, make sure the tutors are doing the right thing if we sort of cure some of the underlying fears about this space. What would be your ideal of what could actually happen if we raise the floor? What could the ceiling look like?
[00:50:11] Dr. Sonia Tiwari: I think it would just emerge from a collaboration between industry and academia.
So to have those two solid pieces, the way we saw in AI Quest, the way we saw in PBS, that kind of ecosystem around how these characters come to be, that's the source of the challenge now, and that's where the inspiration for the future is for me. Typically we had this kind of assumption that, oh, it's an educational game.
It's gonna have like a boring, bright background with like eight different fonts and no one's gonna play it obsessively. We have like this inherent assumption that a game coming out of anything academic or educational is gonna be subpar in quality or it's like gonna be lame. But the more we build these industry, academia kind of partnerships.
The more there's scope for repurposing the power of characters in general towards education.
[00:51:06] Alex Sarlin: Thank you so much for being here with us today. Dr. Sonia Tiwari is a children's media researcher exploring the role of characters as facilitators of children's learning experiences. We really appreciate this conversation and with you being here with us.
On EdTech Insiders. Thanks for listening to this episode of EdTech Insiders. If you like the podcast, remember to rate it and share it with others in the EdTech community. For those who want even more, EdTech Insider, subscribe to the Free EdTech Insiders Newsletter on substack.
