Edtech Insiders

Debating the Future of EdTech: AI Tutors vs. Peer Assessment with Carine Marette of Kritik and Gabi Immelman of Mindjoy

April 08, 2024 Alex Sarlin Season 8
Debating the Future of EdTech: AI Tutors vs. Peer Assessment with Carine Marette of Kritik and Gabi Immelman of Mindjoy
Edtech Insiders
More Info
Edtech Insiders
Debating the Future of EdTech: AI Tutors vs. Peer Assessment with Carine Marette of Kritik and Gabi Immelman of Mindjoy
Apr 08, 2024 Season 8
Alex Sarlin

Send us a Text Message.

Gabi Immelman is the founder and CEO of edtech start-up , Mindjoy, a platform for STEM teachers to 10x their impact for boundless classroom learning through unleashing student curiosity with the power of AI tutors.

With Mindjoy, teachers can transform lessons they love and know using the power of AI to create engaging digital first learning experiences for their students. A former Reggio-educator turned EdTech founder with experience working globally in Africa & USA. Mindjoy is the third company she has founded - her first being a stick-on tattoo parlor at the beach where we used to go for summer vacation when I was a child.

Carine Marette is the co-founder and co-CEO of Kritik, an online peer-to-peer interactive learning platform designed for professors to engage students in a twenty-first-century way. Through a gamified experience, Kritik allows students to develop higher-order thinking skills from creating assignments as well as analyzing and evaluating peer submissions. In addition, students will develop the skills necessary to deliver feedback to their peers through our feedback-on-feedback system. They are using AI and artificial intelligence to increase the reliability of peer assessment and save instructors time. 

Show Notes Transcript

Send us a Text Message.

Gabi Immelman is the founder and CEO of edtech start-up , Mindjoy, a platform for STEM teachers to 10x their impact for boundless classroom learning through unleashing student curiosity with the power of AI tutors.

With Mindjoy, teachers can transform lessons they love and know using the power of AI to create engaging digital first learning experiences for their students. A former Reggio-educator turned EdTech founder with experience working globally in Africa & USA. Mindjoy is the third company she has founded - her first being a stick-on tattoo parlor at the beach where we used to go for summer vacation when I was a child.

Carine Marette is the co-founder and co-CEO of Kritik, an online peer-to-peer interactive learning platform designed for professors to engage students in a twenty-first-century way. Through a gamified experience, Kritik allows students to develop higher-order thinking skills from creating assignments as well as analyzing and evaluating peer submissions. In addition, students will develop the skills necessary to deliver feedback to their peers through our feedback-on-feedback system. They are using AI and artificial intelligence to increase the reliability of peer assessment and save instructors time. 

Alexander Sarlin:

Welcome to Season Eight of Edtech Insiders where we speak to educators, founders, investors, thought leaders and the industry experts who are shaping the global education technology industry. Every week we bring you the week in edtech. important updates from the EdTech field, including news about core technologies and issues we know will influence the sector like artificial intelligence, extended reality, education, politics, and more. We also conduct in depth interviews with a wide variety of Edtech thought leaders, and bring you insights and conversations from ed tech conferences all around the world. Remember to subscribe, follow and tell your ed tech friends about the podcast and to check out the Edtech Insiders substack newsletter. Thanks for being part of the Edtech Insiders community enjoy the show. Welcome to a special episode of Edtech Insiders. We're going to have a debate today between two Edtech founders and CEOs about a pedagogy of online learning and artificial intelligence. Joining us are two amazing guests the visionaries behind Kritik and Mindjoy two EdTech platforms that are redefining educational engagement and personalized learning. Carine Marette is a co founder and CO CEO of Kritik, an online peer to peer interactive learning platform designed for professors to engage students in a 21st century way through a gamified experience critique allows students to develop higher order thinking skills from creating assignments to analyzing and evaluating peer submissions. In addition, students develop skills necessary to deliver feedback to their peers through our feedback on feedback system. They're using AI and artificial intelligence to increase the reliability of peer assessment and to save instructors time. Gabi Immelman is the founder and CEO of Mindjoy, an edtech startup for STEM teachers to 10x their impact for boundless learning by unleashing student curiosity through the power of AI tutors. With mine Joy teachers transform lessons they love and know using AI to create engaging digital first learning experiences for their students. Gabi as a former Reggio educator turned edtech founder with experience working globally in Africa and the US Mindjoy is the third company she was founded her first was a stick on tattoo parlor at the beach that she used to go to for summer vacation as a child. Without further ado, here's our debate between Carine Marette of Kritik and Gabi Immelman of Mindjoy. Welcome to EdTech insiders, we have a really exciting episode today, we have two ed tech founders doing a friendly, very friendly debate about some of the pedagogy of online learning. So without further ado, I want to get in and get everybody introduced and start our debates. So let's start with Gabi Immelman, the founder and CEO of Mindjoy, can you tell us a little bit about your edtech background and what Mindjoy does?

Gabi Immelman:

Yeah. Thanks so much for having me, Alex. It's great to be here with you both. So my joy is a platform for STEM Educators to 10x. They're impactful boundless classroom learning with the power of, you know, AI, specifically AI tutus. And really what we care about is unleashing student curiosity pedagogically, that's something we really care about as the world moves into a place where we want to foster, you know, 21st century skills. And you know, we all talk about the four C's, but curiosity really is the starting point, the sort of flame that we have to kindle to develop young people's motivation and desire to learn. And so that's really what we're trying to do. And we've kind of borrowed this idea. You know, a lot of folks talking in the AI space about teknicks engineers or 100x engineers, we're thinking about how do we build a platform that helps teach us 10x or 100x their impact beyond the traditional four walls of their classrooms that they occupy and physical spaces, and maybe we can reimagine, distributing talented teachers skills beyond just the classrooms they occupy. So that's really what we're about. I'm a former Reggio Reggio Emilia educator. And so really played with progressive pedagogy is you know, centered and student centered learning and project based learning and first school and all these amazing things and learning from our environment and making learning visible. So I've had a lot of experience with that. I'm a South African, and I've been fortunate enough to offer spend time working in Silicon Valley so really had experience working with children from all walks of life from resource rich context to some of the most under resourced contexts in the world. One thing I learned is that children everywhere are curious, you know, it doesn't matter where you're born and talent is really equally distributed but opportunity isn't. And so I think you know, that's a big question on everyone's minds around equity and accessibility. So that's something we care about. I often joke I say that my job is my third company. I used to have a small business as far as exiting the classroom, doing after school programs and that kind of thing. But my first business was tattoo parlor at the beach, when I was a 12 year old kid, I used to try and make extra money in the holidays, because I'm from a family of entrepreneurs that was always modeled to me, I was just sharing before we started this podcast that, you know, when I grew up, I came from a family of matriarchs, and my mom and my aunt and my Booper aunts, and my grandmother, they were all entrepreneurs, and they all started their own businesses. And so I kind of thought this was normal. I thought, This is what you do. So I spent my summer holidays, trying to think of businesses and make some extra money. When I went to university, I kind of realized this was not the lived experience of other young woman. And so I just kind of realized how much models matter. And yeah, when you can, if I had linked that all the way back to AI, tutus, there's a really interesting thing of thinking about, you know, what is modeled for you is what helps form the way you construct the story of yourself, and the opportunities you can access. So I think models matter, excuse the pun on models. So we can maybe think about the society of models that we're trying to create. And you know, that which we scale like, let's hope we only skill, good pedagogy.

Alexander Sarlin:

Yeah, I think it's up to us as a community to scale pedagogy and make sure that we're, as you say, you know, starting with curiosity, and augmenting teacher impact and all of these incredibly important aspects of what Ed Tech does. I'm really excited to you know, let everybody know more about my enjoy. It's really cool platform, Carine Marette is the CEO and founder of Kritik. Can you tell us about your ed tech story and what you're doing with Kritik?

Carine Marette:

Thanks so much for having me, Alex. So great introduction, Gabi. I like it. I have a few notes. I have a few points in common with you. Yes. So getting married, I co founded Kritik in 2018, with more sunshiny who built TypePad in the past. So codec is a online p2p learning platform. We created a platform before COVID. So it used to be an experience in person. But now that COVID happened, everyone assumed that Kinect was created for online which is a funny story to share. So a little bit about me before I go too deep in the platform, so my background is French, I'm not bilingual. I emigrated in Canada in 2013. I have an eclectic background, which is what I like but GAVI as well, so I was a designer for Dior, Chanel. I was a project manager for a lot of different startups in the past, I was an accountant. I work in a farm. I did like a so many jobs that I cannot list today. But it will be very interesting to talk for the fun of it. And right now I'm doing my EDD PhD in insurance system technology at Indiana University. This debate is interesting for me to kind of like talk about learning theory, as well. But I keep that for the end. Gabby, I like when you make a draw when you say oh is your third company? Well, me to guess what I created a startup before in fashion industry to build clothes customized and at that time, I wanted to use AI. But I did not because I didn't have the technology. But I was thinking okay, what could I do if I could create a company where people are so obsessed with like those fashion icons Oh, I love that clothes I want to buy but they don't have the money to buy that same clothes for like 1000s of 1000s of money, why we can just like give them an alternative using AI a product picks you and then you say I can buy the same cheap clothes to another local shop here or anything from you know, important shop. Anyway, that was an idea that I had, I ended up being like a dressmaker back to my first job. And I stayed for a while there. And then my second company was a smart insole. So I created Fitbit, but in the shoes. So my friends were like PhD students. And then they helped me to design the powder and use math lab and algorithm to kind of get our mind like how people work. And then the application of it was to help senior even error to the doctor if the senior fall because there is no weight on the insole. And then also yoga application or sport application. Like I have my Garmin now I don't know if you have I have this kind of like watch. I think people are obsessed with data these days. And then of course the last one is critic now. You talk about your family like having a model Gabby and then really like that. And then one more time. I got something in common with you because my dad was entrepreneur. My granddad was entrepreneur and my blogger family like had a lineage of entrepreneur. And then when he came to me I was like oh my god, I have no choice. I want to be on destiny. And then funny story is like that does not have son. So for him as well is like okay, Karen is going to be the entrepreneur Are you a legacy? Well, interesting story, he said, I cannot continue the legacy, you will be the entrepreneur to my company, the next one. So he wanted to take the same businesses and know what I'm going to put in the legacy, but in the area that fits my belief, where I want to stay stable in Alderman. And then my mom's draw was be the Buddha version of yourself. And then when I had the sentence in mind, I had that, okay, I want to pass myself, I want to have a bit of skills, I want to do like a lot of different things. I was very passionate about it. So when I came to Canada, my English was very poor. So I went back to school to learn on my I had a master's degree in computer science back then in France. But then when I came back to Canada to go do my kids, well, believe me, I have to start from scratch. But it was good, because it forced me to learn faster. And then I said, Okay, I want to continue, what can I do, I decided to do a bachelor degree in psychology. So when I was doing that, my goal was like, okay, you know what, I want to have a plus everywhere because I got my master with a minus a b plus. And I said, it's not enough, I want to do more, because I'm an adult now. So So and then I was frustrated, because my professor did not kind of like, give me the feedback. And I wanted. And this is how the whole journey of kick started. I wanted to have a system. And then I created an online book, on my google doc to share information, but then no one were doing it because they were afraid of cheating. So I say it's time, it's time to create something. So Kitty came to mind. So I wanted to focus on developing higher order thinking skills, and analyzing and evaluating pure submission in order to learn. So you, Marcia, and Gabby, the term curiosity, you are focused on learning curiosity. For me, my focus was like critical thinking skills, like I want to really to just develop that. And then we focus the most on delivering feedback to peers through a feedback feedback system. And there's some gamification into it. And lastly, just to end, my introduction, is there is some AI as well and artificial intelligence that you use to increase the reliability of peer assessment and save instructional time, you

Alexander Sarlin:

have so many interesting things in common and one of the things that strikes me You know, I think you both mentioned that you sort of bridge different kinds of worlds, right, you know, from the tattoo parlor on the beach, and, you know, working on a farm to Silicon Valley, you know, work and doing project management, and doing fashion startups and all of these really, very high tech skills or, you know, insoles, smart insoles. And I think that kind of experience gives you both a really wide reach in sort of understanding your learners, your the educators you work with, and the technology that can be used to really enhance all of their work. So before we jump into the questions, we just want to do a couple of really quick definitions, because this is something that, you know, I think everybody will benefit by understanding your particular definitions of these terms. Carine let me start with you. You mentioned that critique is a peer assessment platform. Can you just tell us a little bit about peer assessment versus self assessment and why they're different than sort of traditional assessment?

Carine Marette:

Yeah, thanks for asking, I think is great to define. So peer assessment is when students take on the role of assessor to evaluate and provide feedback on the quality of work produced by their peers. You know, it can be based on the on the rubric designed by the instructor, and clinic, we often rely on that to make sure it's very objective, because there's a lot of assignment that could be subjective, but when you have a heartbreaks, it helps. So this method, leverage, like the diversity of students perspective, Anchorage collaborative learning environment and develop critical thinking skills. And then on the other hand, this self assessment is where students evaluate their own work promoting self reflection, self regulation, and a deeper sense of ownership of their learning. So maybe what I could share, to be more in depth is like how it works in clinic. So there's three stages. So there's first the creation stage, which is called stage one. This is where the students create and submit the creation. And there's a stage two which is a valuation stage. And this is where students analyze and evaluate peers creation, and as well as their own. So during that evaluation stage, we can have the P assessment as well as the self assessment. And then lastly, the stage three, which is called feedback stage is where students provide feedback on the evaluation they received. So in this context, we rely on students and algorithm to provide fair grading and happy to hear more from Gabe as well.

Alexander Sarlin:

No, that's really helpful and Gabby you mentioned AI tutor is we're in this amazing age where this AI tutors are starting to come about tell us what you mean when you talking about AI tutors and how they provide sort of personalized learning or how they might support inquiry based learning in the sort of curiosity that you're focusing on. Yeah, sure. So

Gabi Immelman:

I think it comes to air IQ does it's not a new idea, right. So the idea of looms two sigma problem is something a lot of, you know, I get for everyone in the mastery based learning world and the platform will this is something we've thought about for a long time and Khan Academy's great example of an incredible mastery based learning system that seeks to find an answer to blooms two sigma problem, which I often like to summarize, as, you know, the question that poses how do we make group instruction as effective as one on one tutoring? And currently, you may say, well, peer assessment is one solution to that problem. Where we play is thinking about basically unbundling expert teachers and asking how do we create enabling environments that enable young people to access expert knowledge. And Jenny, I, you know, has totally changed the landscape compared to what we used to have in terms of old, you know, machine learning systems, or machine based or adaptive learning systems, which are incredibly, I'd say rigid, and compared to what the world of Gen AI or generative tutors possibly present an opportunity for almost as flexible, a adaptive experience for students a student led experience that is, you know, feels more like talking to an actual person. And so from that perspective, yes, you can have embedded pedagogy or learning outcomes, but the pathways are completely undetermined. And so the opportunity that I think is incredibly exciting in this new world of personalization, it really flips this idea, and that young people have to actually start being agents that drive their own learning. And it's kind of meta, because as you know, in the in the AI space, if you start looking at, like the agentic models, and said, you know, infrastructure for getting agents to do things on behalf of students, I think there's a really interesting fundamental tension that we're, we've never seen before, which is, yeah, we want to outsource things we don't want to do to, you know, AI systems, but learning requires us to do things that we don't want to do to really grow, you know, sort of proximal development, all of that. So there's this really interesting new space. And for us, we care a lot about inquiry based learning. And the Socratic method, as one way to not have a system that just give students answers, but actually engage students and bring them in and get them to think critically. So often, we have teachers tell us, our students was so frustrated using my joy today. We loved it, they had to think. So, you know, obviously, we want it to be a good student experience, too. But it is that tension of like, this is hard. But it is also fun and meaningful. And to still borrow a term from a great computer scientists and educational technologists, we often talk about hard fun at my enjoy. Seymour Papert was one of the founders of the Media Lab at MIT, and was really, you know, a pioneer in terms of computer science for children creating logo, but also a great mind and trying to distill how AI systems could work taking inspiration from how children learn. Fun fact, he was also South African. So he has this idea of hard fun. And that's the pleasure of like, almost like the Fineman thing, the pleasure of finding things out the finding the challenge enjoyment and the challenge. And so that's something we're trying to

Alexander Sarlin:

inculcate. Yeah, you don't want to outsource the critical thinking to you, ah, you know, you want to make sure that if it if you have an agent doing some of the maybe the research or the lookup for you, it doesn't mean that it, you know, it should do the thinking for you. Fantastic definitions. And so, with that, let's jump into our debate questions. And I think this is a very natural segue here. So our first question is, in the context of the type of inquiry based learning that you just named Gabby, and your Socratic as well, which method is more effective? Is it more effective to engage with AI tutors to ask questions, build curiosity interact? Or is it more effective to collaborate and evaluate with human peers and work directly with other you know, real minds? Carine? Let me ask you to kick off our debate. How would you answer this question?

Carine Marette:

Thanks. So this is a great patient. Just before I answer your question, I just want to highlight something that Gabby Marsha, which I really appreciate, because she does very well on it, she motion the blooms two sigma problem. And that's a well known phenomenon is basically for people who don't know is like, you tutor one on one students, with the professors. And then you try to master the learning technique, the skills. And I thought long time ago, it was a problem that we couldn't scale that so. And then what I appreciate with the technology of Gabby is like she's now scaling with using AI tutoring that phenomenon. And I think that's very smart of her. So now to answer your question, Alex, so the question about inquiry based learning. So pm assessment has been around for a while, and then some people didn't have access to our technology gap your mind. So they relied on multiple choice questions. And then he was like, yes or no, they were no, too much. No much inquiry on that because it was like they wanted to they couldn't scale the problem based learning. They had to have yes, no answer. So now you have the technology. I think academia is like providing Higher Logic Asian some Kritik, and that sense is engage with peer assessment, engage learning and provide a quality and consistency in feedback. So we promote critical thinking skills, engagements, skills development, and you know how to effectively address peer learning viability challenge. So I'm going to have a motion maybe three points to answer this question. The first one is like or incorrect, we have a online GAVI friendly debate, we do not rely on AI, we really emphasize that there is an importance in the human element. And the logic behind it is we use the Garner stare theory of multiple intelligence, where human peer provide empathy and emotional intelligence in enhancing soft skills development. So not only we helped to develop them the soft skills, but also we can have like this perspective of human develop, like a sense of responsibility, the sense of belonging, and then we help them to kind of like make connection and provide viewpoint in like the development of their soft skills like how then okay, sometime, you know, agile, we have seen a lot people reacting, oh, what are you saying that and it was like really, like, be upset. So, like, it's kind of like close to behavioral therapy, where all behavioral perspective where we say, Hey, do not react, try to step back. And it's like, you know, provide feedback. And then so for that we have values features, you know, we assign evaluation, we have a group activity that I can talk longer later about that there's different type of group and live presentation. So now, the second argument that I would like to motion is like, there is the depth of understanding. So I don't know how gave me those views of AI, I suspect that there is like pre loaded data where they have to do something and then, so I'm going to talk about mine, and maybe you can expand, but we do he actually focus on learning by teaching, we help students to, you know, develop diverse explanation and viewpoint, it has to be really objective, there is a whole breaks, and then there is a structure of assessment. So one thing that I want to share is like, you know, that you are doing a good job when beyond the classroom, the students share their knowledge. And what happened in a classroom of like, I think it was about like 1600 students, one students went on Reddit, and then say, Hey, I don't understand the materials about I don't understand how to learn about that topic. And the students say, Hey, I use correct and I add the blob, I created a video and I teach my PA using that, this is how I can help you. So quickly, the students up to one step further, say I can not only teach my classroom, but I can also go out in the web, and then share my knowledge, it's kind of like the value of having a portfolio where you develop your skills, and then keep that for your learning. And then the last point that I want to share is adaptivity, I don't know how Gabby could adapt to the social learning, because AI is not, of course, AI is working with the students, or maybe I'm looking forward to hear from her. But um, I can say for me is like, this human interaction in a learning environment, you know, flexible, is like there is like variability in in the response. And then it's also based on the group. So sometimes the group understanding and what kind of issue they want to fix. And then just to be short on the grouping is like, it can be based on preference, it can be based on like, talent or different things. So there, they can really go in depth in that adapt, and then, you know, remove unnecessary feedback. But where adaptability works, the best for us is something that's called calibration. So the calibration is like, is different than a peer evaluation, there is no three stages is only the second stage where these are evaluation only. So what happened is like, it's the timeline by the instructor, the instructor, they say, Okay, this creation value these grades, and students need to determine, Okay, what is great that I have to give, and then when you do that, there's two purpose for that one is the algorithm behind the scene will look at the grade given by the students compare that was given by the integer and then the delta will the time I have a grading power of the students. And that grading power will be used as a score that determines that critical thinking skills. So therefore, it cannot lead to accuracy of grade and I'm looking for for Gabby to challenge me on that. Well, if I have a high critical thinking skills, then when I read my peers, my score will count six times or more. But if I have low critical thinking skills, I am included in a peer assessment, but not to expand for other peers to be penalized for my grade. And of course, the second purpose of calibration is determining the prior knowledge. So the outcome of that calibration the professor can filter and then we can say okay, this one understood. That means you have to prioritize so it helps Well grouping as well. What

Alexander Sarlin:

I'm hearing from you Korean is three points about for peer assessment, one, it builds the soft skills and the empathy and the sort of different types of intelligence to have students working with each other. Another is that you can learn by teaching and students can actually, you know, develop their own understanding, as well as develop artifacts by creating teaching material for each other. And the third is this idea of adaptability, and calibration, where you can actually help students at different levels, contribute to each other's, you know, grades in different ways, and help people who are maybe behind in critical thinking enhance their critical thinking skills. Over time, I'm thinking I'm paraphrasing that decently, but maybe not struggling with some things. But, Gabby, let's put it to you. So in the context of inquiry based learning which method is more effective AI tutors or peer assessment? Well,

Gabi Immelman:

I think it very much depends on a certifier to kind of have a critique of peer to peer learning is that one of your assumptions in the efficacy of peer learning is that the knowledge of the group is required to understand and value and know how to facilitate that kind of experience. And so where you might find that's really, really difficult is for younger students, say in K 12, that I've never participated in environments like that way, that's not how your family or your school environment or any of that is set up, you still have very, very traditional rigid understanding of how knowledge is built, it's very much still that I go to school, I answer questions I write test. And there is not a willingness to exchange ideas or put ideas forth. And actually, that is a really, I almost want to say an advantageous environment to be come from is where you know how to do that, where that's been modeled for you back to the idea of models. So I think it does have a concern of tremendous value to even something like peer feedback, is in environments where we don't have that knowledge, that knowledge is not inherent in the teaching child, or the pedagogy or the family or community structures, and where we can create things like simulations as an example. So we, for example, our platform builds infrastructure, and teachers can build their own tools, their own tutors that are fit for purpose. So for their classroom, their lesson goals, they can also learn from expert teachers across the globe, who are better than them at building these tools or tutors. And they can use, you know, it's always a marketplace of tutors that I can borrow somebody else's tutor, I can folk a copy like you do in programming Folker copy of a file, and modify it and remix for your purpose. And so so we enable that. And what we've done is we've created a lot of templates, sort of research line templates, based on Ethan Pollock's work, where he's got this great initial study on how AI can be used in the classroom. And so we've actually built tutors aligned to that research that teachers can use and modify for their classrooms. And there you see a variety of really interesting use cases or AI tutor as a coach. So where tutors are actually helped students go through maybe some social emotional learning or soft skills, experiences, you know, dealing with a tough time, we've seen teachers do amazing things like a cyber bully coach, as an example, is something one of our teachers has created, that's actually created a platform for students to share their experiences. And because there's a human in the loop, and there's supervision, you know, if there's students that need support, we have a moderation system that helps give teachers insights, and help them identify any students that are at risk or need to be supported with something. So that's one way we help with social emotional learning. It also serves serves as a support to teachers who might not always know how to facilitate those conversations. So that's where something like a simulator as a simulation, a tutor that helps you simulate an experience we've seen teachers do all kinds of creative things with simulations, you know, and then you've got just like revision, like Cheetos, or academic coach, or like a traditional tutor, and that way helped me prepare for an exam, practicing questions, that kind of thing. So, again, you know, there's a variety of different use cases, depending on your, you know, your dynamics of your classroom, and where your classroom situated, and what the knowledge is of the community. But I think what's really exciting is that tutors can embed certain, you know, pedagogies, like, for example, we default Socratic mode. And we've trained 1000s of teachers to learn what Socratic method is, you know, because there's, it's in the UX. So to have, what is the Socratic method? What does it mean, what does it June, we can actually then the product becomes a teaching moment. And they're like, Wow, that's a really powerful pedagogy. And so we're trying to think about, like building products in that way. And I think chatbots are just here, the first evolution of of that, envision meant that Yeah, I think it really depends on where your students are, what their level is, and the knowledge of the community. And so I think, where AI tutors are incredibly powerful is when you don't have access to that knowledge and where you can distribute knowledge, not just information, but actually even more Do behaviors or scenarios or practice situations? So I think really the possibilities are endless.

Carine Marette:

Thanks so much. I think I have no choice to rebut that. I have a viewpoint, I think, maybe three, four. So the first one, I say, you say, okay, the knowledge of the group like you can have like that doubt about that. So what I would say is like, in the system that we have, let's say, if the knowledge share either on evaluation or any feedback, there's always this human perspective where those students can dispute say, hey, I want to dispute this fair, that is great that I receive or this explanation that I got is not fair. And then they dispute and the prosecutor the professor can intervene and and resolve a dispute. And then he can either remove the grade, say, Okay, this is not going to participate, this is not acceptable, or I get it. And then when the professor edit, what happen is the AI behind the scene will make micro calibrate. So it's gonna save the calibration, but within the activity, where the professor will say, okay, that does not deserve five stars, it will deserve like three stars, and the delta will impact the whole pool, and adjust the critical thinking skills, oh, the grading score? And then well, let's say it's fine. What can we do more, what we could do more on top of that is we have something like a collaboration to help with knowledge. So we can add a subject matter expert. And then those can provide feedback. So that's very interesting. Now, when they provide feedback, and that depends on the policy of the professor, you can always provide feedback before the creation stage is ending, or after. And it's also like an evaluation note where the notes can be preloaded, because you talk to me about like, okay, it's difficult for junior participants. What about those who cannot do it? Well, for those who are skeptical and afraid of it, there is an option where the professor can say, okay, in one hour, I'm going to give them the solution. Because I clinic is not about finding the right solution. It's about I say, Okay, here's a solution. Now, what is the thought process for you to get there, and then they have to explain to develop their soft skills. So it's not contagious, to take your words gave me is know about like, is very unleashing it enriched the learning. And then we help them train them to provide meaningful feedback. And more importantly, is, it's kind of like anonymous, where they feel safe to participate. And then you talk about like the revision. That's very interesting, because the goal of peer assessment is like, also, when I visit your mate, the first creation, you get the group of those three chairs, three stages, creation, evaluation, feedback. And by the end, what do you do? Well, you analyze your learning activity, or you just say, okay, that's now my final document where I include the feedback was it? So now, maybe back to you gave me if you don't mind Alex asking her questions? Well, I think that in my perspective, there is a limited depth in understanding in like aI tutoring. And let me explain you why. Because I think that the data train from AI is a scan of like, predefined information. And then it could be bios, and then it's very narrow. And it does not kind of include the contextual nuances that human interaction capture. So I'm saying like, for me, my perspective is a little bit more superficial understanding, and then it doesn't it does not engage in depth. So what's your thoughts on it? And also, the second thought is, how do you do for teamwork? Do you have anything about like, what is teens? Yeah, sure.

Gabi Immelman:

I guess there are a couple of things. So I think the limitation of knowledge is a technical one, I think more and more systems are starting to enable tutoring systems to embed corpuses of knowledge and leverage, like curated high quality corpuses of knowledge, so often, something that we see so for example, if you have a teacher in one context that teaches IDCs e, that will, the Cambridge curriculum versus the international American curriculum, you can allow for basically, selection of curriculums and large corpuses of knowledge has to be leveraged by AI system. So I think that's just the technical one, and what we're solving for sure, you know, you can basically leverage our system to also tailor responses, giving the system context awareness of the students. So let me give you an example. If you set up a tutor, you can provide them with context of your classroom, the students you work with specifically so you say these are students that live in a rural place, a rural farm town in South Africa, versus these are new york city children. And I would like you to explain this concept to them. And the system can deal with that and give context relevant examples so you can even use Show and Tell listen to example to be like. So an example would be, make sure your explanations include local currency and the way you prompt a tutor, you can make sure that like, it's astounding sometimes what the references are. So how much awareness actually, if focused in the right way. And if you provide the system with the right context, how much personalization and context relevance you can actually create for learners. And the ability to deal with ambiguity, I would say is another really interesting thing. So if a student or a learner says, I do not know what that means, an example would be maths, giving a maths problem with like, Alright, so the system says, candy, like your candy, say you wrote a terrible prompt in your tutu, it really, really struggled with anything. So one of the amazing things is you just need to kind of know to go, I don't know what that means. I don't know what a candy is. And they'll be like, Oh, it means chocolates. And you'll be like, I don't have chocolate. Like, I don't really get chocolates, where I'm like, on top, it's not really a big thing where I'm fun. And then you say, alright, well, like, what do you have a lot of like, that's a treat. And then you might say, Oh, we have mangoes, and lots of mangoes, the fruit, and all that. Okay, great. So let's go with mangoes. Then if you had a mango, and you cut it in half, like blah, blah, blah, and then all of a sudden, that live on biggity is a very, very different form of personalization that is possible versus an adaptive learning system as an example. So that would be sort of my rebuttal. In terms of fist sizing, I think, like, what we try and do is we obviously try and engineer away bias and safety and harm, and we looking at all of those kinds of things, because we want to use AI for good. And so that's where human in the loop matters a lot. But I also think that one thing we care a lot about is talking about, making sure and then this is maybe an African sentiment is a lot of people like Earth's by a stern to use the systems Well, if you are not a customer, then your values and the things that you stand for will not be represented. And so participating, being an active consumer and voting with your dollar and being represented with your dollar matters. That's something we advocate for in a big way, which is empower yourself to know AI literacy is one part of it, but also empower yourself to be a user that gets represented fundamentally. So that's something we think a lot about when we talk to teachers. And, you know, we care a lot about also just having these conversations and and creating forums for teachers and folks from all over the world.

Carine Marette:

Thanks for answering I think like I like the fact that, you know, you define that, how it's not as superficial as it should be. But maybe the second part of my question was about, like, do you have anything to work with team I think like when because we do emphasize with like teamwork or things like that. At Kotick, we have two type of groupings that between groups or within groups, this is how we engage teams in our platform. Sorry, I missed that was a between groups. So between groups is there is interaction between the evaluation stage. So as I told you before, there's three stages in codec creation, evaluation and feedback. So in the creation stage, each group will submit one creation group a submit one group visa and one and then during the evaluation, the group as we evaluate V, and then the same for the group B will evaluate Group A. So this is basically how it works. Now, professor told us like, okay, the additional way of doing group work, when there is one submission, you know, you don't know if those participants participate. So with that type of group between groups, you we know that it's efficient, because during the evaluation stage, they will participate and they will need to think critically, despite having one question submitted for a group, everyone will participate. During this activity P activity of between groups, some professor said sure, I like that, but I would like to have everyone participate in the creation stage. So that's how we created the next activity type later on, which is called within group. So within group is everyone in a group group A everyone a group A will submit vibration, and one group B will submit vibration, and they will evaluate each others. So there is no interaction between groups, the group is predefined by the professor or predefine via the algorithm. So how to use the algorithm for that. So we have the calibration features where the professor upload the tree creation, the score is given by the professor. And then the students need to evaluate those ration and the process their score is to the professor the higher the critical thinking skills will be the score, and then from the result of that activity, then the professor can rank those students and then from the ranking they can say, Okay, I will group the students with the same level. So I've taught that I take the top 10% And then a group then this and then so on, so forth, or I will take the I will have to have a mixed match of levels. So I will take some students from the beginning and then some students for the, at the end. And then so it depends on the philosophy of the professors. But what the point is in Kotick, we can identify prior knowledge via these calibration features. And then now we can group students based on their interests, experience, knowledge competency, and that's basically the value that we are giving it here. So another thing that we could do as well, but some professor I've seen had done is they use personality test to group students, and then from the personality test, they will know who can be matched with another. So that could be part of the learning activity, where the first critical activity or a survey of the procedures need to be completed. And then they need to, they need to reflect on it. And then from that activity, then the professor can group the next students into the very insightful collaborative projects. And then one thing that we should not forget is we are a peer assessment platform, but also there is a self evaluation as well in those coping features. So this is where he works. So I would like to learn more from you as well. Like if AI tutoring is missing this element of like inquiry, because we are in the topic of inquiry based. So do you have Yes, sir,

Gabi Immelman:

again, like so simulations, for example, are game based or team based, often team based experiences. So you might have one tutor facilitating group work or conversation. Another modality that we use a lot of is like something like encouraging good pedagogy. So or something like very simple activity, scaling an idea, like think pair share, as an example. So that is not about just one on one with a tutor that is about work with the tutor, then pair up with your teammate, work with your teammates, and participate in a big caution discussion. And so what you often see, and a lot of the feedback we get, as teachers saying, Wow, this is enabling students whose voices I never previously heard to participate in conversations because they had a chance to practice ideal, or work at so you know, those ideas where you can scale where you have, yeah, you facilitate peer work and group work and teamwork. And then just basically, multiplayer chat is the roadmap.

Carine Marette:

I like what you're saying. And I think I want to also say something like, I like this motion, previously, and now the power of like, active learning, and you know, wow, I never heard about that before. So and also just thanks for publicizing, like how those dynamic works, because he helped me understand the value of your platform. So what I wanted to add here is we do also value in the context of like, irrelevant example. Like one thing we have is like, multi topic. So we use like, kind of like in the creation stage and ability of multi topic. So the students can have be assigned to a topic that they like to debate what you need to discuss, or it can be random. And then basically, what happened is like, the professor will use the same learning activity to Sam or breaks. And it's kind of like almost like, like competency based education, where the application of the learning and multi topic can be either lower or and a higher level. And then those who have the prior knowledge can apply in depth, their learning, and those who are just beginner can just still build on their knowledge, as well as what you mentioned before, you know, you said different topic completely irrelevant. So you can say the same topic, but one in France, one in America, we could also do that as well. And then oh, it's maybe it's getting too long for that topic. But I still have like we do. Like you mentioned the term ambiguity. And I think that's a very nice term to highlight. Because in Quebec, we do have like a structure and assessment, like to remove this ambiguity. Thanks. Thanks for letting me like asking those

Alexander Sarlin:

questions. Fascinating. So let me just put some of the things that we've heard it's really interesting debate. And let me paraphrase just for a second, just because there's a lot going on here. So Gabi, you mentioned that you have simulations that AI tutors can actually model different pedagogies like Socratic Method and teach teachers to do that, but that it also the tutors can model thinking to students who may not have the prior knowledge or the context to be able to perform peer assessment effectively. That was part of your points. And then Korean, you're saying that AI can contain bias, maybe that it doesn't have deep enough knowledge to provide context. And Gabby, you're, on the other hand, you're saying that AI tutors can actually be prompted and designed to include context or through interaction, it can learn the context of the student environment, like the mango chocolate example. And then lastly, we've talked a little bit about group work, which I think is really a very important part of this debate, right, which is, you know, peer assessment is sort of designed to be social. It's designed to be group oriented and can work with lots of Different types of assignments including group assignments. And is that yet true for AI tutoring. And I think Gabby, you're saying it is beginning to be true. If there are some definitely some ways that group work can work with AI tutors. as facilitators for group work, you already have social pieces that way you model think pair share and other social pedagogies. And that this sort of multiplayer mode type of thing is coming. And then you're seeing curried as well that this add activity, the idea of being able to adapt with these sort of multi topic type of work is also a really core focus of peer assessment that you can use the human in the loop and the teacher prompt, as well as the sort of systems understanding to be able to tailor and personalize the same way that AI tutors sort of naturally personalized, a lot of really interesting things going on here. So I want to ask about another incredibly important aspect. Education Technology is all about scalable education, that is a huge part of what it does. So in terms of peer assessment versus AI tutoring, which type of platform offers greater scalability and accessibility to diverse learning communities that may not have access to education in this way? Is it AI tutoring systems that are 24/7 that are accessible, you know, to students on their devices? And global? Is it pure assessment platforms that are able to expand the reach of any educator to include a huge variety of different evaluators and people who give feedback, including on systems like Reddit, like you mentioned, Kareem, Gabi, let me start with you, which platform offers greater scalability and accessibility? So

Gabi Immelman:

I think there's room for both. And that's maybe a little bit of a cop out as an answer. And I guess it just depends on how one builds one's platform. And that kind of thing. One thing we spoke about at the beginning was, you know, social learning. And I think one of the most interesting phenomena fundamentally, is that learning is basically a social phenomenon. We learn way more from each other than we do, like sitting rote studying, we're just actually constantly learning. And I think one of the things we think a lot about my joy is, and we often try and fight the idea of like, traditional tutoring, and in many ways is, the thing that we actually care a lot about is the importance of motivation. And I think social learning and connection is so so important in today's age, right? It's essential, like young people are lonelier than they ever were before. They're all these kinds of questions. And I think the value of shared experience has increased. And the importance of shared experiences increased. And so I don't think classrooms I think if the pandemic taught us anything says that questions are not going to go away for most of the world. Yes, micro schools are a growing trend. And there are all these alternative pathways. But for most of the world, if you look at the African continent, most of Africa is going to still be educated in traditional school settings. And those are fundamentally social learning experiences, the traditional classroom is essential is an opportunity to cultivate social learning, but very few classrooms operate that way. And so when I talk about scaling, good pedagogy, we care about changing the way those classrooms work, we want to transform from the inside up, because kids getting together that school is a great thing. There's wisdom in the old, but there's also a need to do things differently. And so I think the scalability of AI platforms presents a tremendous opportunity in facilitating that change. I think it's going to happen much slower than we think. But I think it's going to be more drastic than we anticipate. So I think that's maybe a short answer to your question. I think the other thing that we think a lot about is this idea that resource rich contexts will always have master teachers, I don't think teachers are going away great teachers will serve at the top institutions. That's sort of the law of excellence, like the top teachers go to the top. But then the question becomes the deficit is really in verbal classrooms, classrooms and poverty classrooms where there's a lack of resources, and what do we do in those spaces? Because we don't have enough people, you know, we have by 2030, will have a 26 million teacher shortfall. And I always say like, Oh, it's so interesting that 26 million software engineers, so if you're teaching things like STEM skills, scaling, you know, engineers, getting engineers to give all the time and technical students and computer science grads is not the way to scale STEM skills. And so, which that's specifically what we care about. So we asked, like, how do you take a non technical teacher and empower them to teach technical subjects like science and coding and mathematics? And we think, you know, we did some really early, interesting collaborations with a company like red blood in the early days with mobile coding, and what was in the start of, we were part of the check up t beta as GPT threes better and looking at how do we take a difficult subject like teaching coding, and how do we empower non technical folks to teach it in environments where you don't have Chromebooks, but maybe mobile first environments, and what was really exciting to see is, oh, how all of a sudden, if you don't have the skills, you need to teach us asking questions, being able to follow up with your questions, this idea of curiosity. But if you know how to do that, you can guide your curiosity anywhere. And so that is the tremendous excited thing for me about the scalability and accessibility is being able to take expert knowledge, basically, the cost of intelligence, right, is going down, and we're able to scale it to places in the world where people wouldn't be able to access that kind of knowledge. And that is an incredibly exciting opportunity. Yeah,

Carine Marette:

I couldn't agree more. I think for me, it's like, you know, we are here to transform the school education. And I like that KB, emotion, the role of the teacher. So for us, do you answer the scalability equations, the role of the teacher is important, because we want to kind of like, empower the students to take ownership of their learning is like the same thing, Gabe, you say, like, we want them to be curious, we want to be own and and we want that help to be I'm sure that the other clients at the end of the day, they pay the education, right. So just like own it, and then develop the skills that you need for the workplace. So the role of the teacher here is about not being any more teaching. But on top of teaching, they need to be a facilitator, where they facilitate knowledge between students, and then they basically observe them. And then of course, to guide them, like I like this word as well like giving them feedback, correcting the direction they are reading the guide. And then I like the word that gave me say, like in terms of the lack of, or sources, or the like, everything, I will add on top of the deck the gap of knowledge between students like how because at the end of the day, we want to help all students to learn together. And it's kind of like quite difficult if you are not in the world of scaling, right? Like, how do you scale a mass amount of students like I'd say, 1000. So to answer this example, I will just give an example from Africa as a Gabby's from there. So we have a client from Africa. And then this university has like 7000s, of African engineer students. And then basically, the role is to teach tech subject the same way that you mentioned, or your gaming. And then basically, we help them to develop a good communication skills on the tech subject. And then they are aiming to join the company that we have worked before gateway, like something located in Silicon Valley. So that's the aim. And then we are them there to engage them and enhance their communication skills. So here in terms of scaling that process, we help them to give and receive authentic feedback among those 7000 students. And then we make sure that there is an anonymous part of it, where they will be free to participate. And I have another search, which is like a professor Valerie wellborn, from Virginia Tech. So basically, she was always Tech students. And then before when she was in class, the minority such as female engineer, didn't want to participate, because they were intimidated. And then so what happened is like with critics, you find out that not only they participated, but also they have like a very strong communication skills in providing feedback. So that was a great discovery. And then they became more engaged i and then and she, this is how she skill, like authentic feedback using Kelly. And another example will be David Wong from University of Waterloo. He also like conducted the research. And then he also highlighted like how the formula participated as being a great leader, and then removed the lack of confidence in those things. And then lastly, when you talk about scaling, I would like to say that it's important to scale so for students with a disability, so we are copying to the like Web Content Accessibility Guidelines WCAG. And also we do late submission for students with disabilities. So they have to, they can submit late, but they have to specify or write something down for the professor accessible to Professor only. And then now the professor instead of checking email and things like that. And now we can scale that process among them. And then something really on top of it is we have live presentation where one students can present in a classroom, that presentation is recorded or uploaded in clinic. And now students who cannot just like see or are not in class or things like that, they have access, and then they can just do ask that kind of feedback and then basically that helps to adapt to a variety of mediums. So we are supported like we are trying to do as much as possible supporting the principle of universal design for learning UDL to support diverse learning, different type of material, text, video image live presentation like today.

Gabi Immelman:

I guess in our world, avid multimodal is really, really exciting. Just offer to increase interactivity. from not just you know, you can work with various different types of inputs. So visual audio, texts and tech staff are sinuous up really, really exciting space. So we also build with all the accessibility web standards in place, and really try to leverage for us we build mobile first, just because we're bullish on mobile. And, you know, it's just really interesting to see how second nature mobile is to young people. You know, I think sometimes it's hard to empathize. If you're even a millennial, with young people growing up today, that mobile phones are just really an extension of them. And you think, Oh, you couldn't possibly type on a phone like that. And yet, it's second nature. And so it's a different form of adaptability. But our earn adaptability is actually sometimes quite surprising. And I think counterintuitive, too.

Alexander Sarlin:

Yeah, lots of great points here, again, just a quick synopsis, because I think there's such interesting stuff going by, and lots of points being delivered in a really short period. So Gabby, you started by talking about motivation, and how having that shared experience is really, really key to getting people excited and motivated to build on their curiosity and learn, and specifically sort of honed in on how resource poor environments that don't necessarily have the highest quality teaching can use AI tutors to enhance their teaching without needing a lot of additional context. So that was a sort of vote for the scalability of AI. You also mentioned mobile first as a way to provide access and this multimodal approach that parallels what is being happening in UDL, with peer assessment, you both mentioned sort of the value of elevating student voices that aren't always heard in the classroom. And I think both peer assessment nai tutoring can allow people to engage with students who may not otherwise have been able to Korean, you talked about the teacher role, which I think was really interesting that, you know, in a peer assessment environment, the teacher still has a very strong role in sort of structuring how the facilitation takes place, which may or may not be as true in an AI environment, I think it remains to be seen. You also talked about the gap of knowledge and how you know, in environments where there is resource poor education, peer assessment can actually support that kind of environment by creating a framework where students can teach each other and build on each other's knowledge, including in anonymous environments, you both talked about disability a little bit, and the value of STEM and how difficult it is to get technical educators into classrooms. But how both of your methods can supplement and support that type of teaching. Very good points on both sides. And I think a lot of cases to be made for the scalability and accessibility for both technologies.

Gabi Immelman:

So we're busy with pilots, which is kind of interesting. And our theory on impact from just a teaching and learning perspective on curiosity, specifically, is based on question handling. And so one of the things we're seeing is, it does provide teachers with the ability to deal with more questions in the same period of time as that they wouldn't be able to So quite simply, like a teacher who can answer on average, and like this is sort of an average thing, which is like, teachers, on average asked about two questions a minute. So what's interesting is they asked two questions, and the habit that is built, students answering. And so what we're seeing is two things, one, and maybe 10 questions, 1020 questions, if you have a very active teacher, who is not just doing a direct instruction type, scripted lesson, but is facilitating conversation is able to answer 1015 questions. And so what we're not saying is that say, for example, in to even a traditional classroom, where you use sort of an AI assisted thing per share model, all of a sudden, you have 30 students, 40 students, in some cases, 60 students in a class, and one teacher have the ability to ask 3000 questions in the same amount of time. And what is interesting is that they're not just especially with a Socratic method, for example, as a pedagogy, they're not just answering questions, they're also asking questions. And so from the skills development, you're seeing the development of asking, and answering questions, which we think's really, really exciting in terms of inquiry based learning and developing and children's capacity. But I think that's something really, really exciting in terms of what it means to be 100x Teacher. Yeah,

Carine Marette:

and then I guess the last thing that you know, I like that gave me emotional about piloting those things. We do also pilot things, and I think that I didn't mention during these debates is I only talked about the calibration and make micro calibration but we have more things like we have aI generated content, where the professor can upload the syllabus, and from that syllabus, the whole course can be created the learning activity, the whole breaks. And then for students side when the students sometimes say Okay, now what about is Students use AI and then generate content. So we have like aI plagiarism detection, where the professor can click the button and then see what the likelihood of that detection. But I'm not going to debate here. I know we can maybe keep that for another time because don't have time. But there is some flowers with a lot of detection. I think it's a good debate, maybe if one was a time when I wanted to say back to now, the purpose for me of this debate, as I'm students and doctoral students at Indiana University, I'd like to end this debate that we, throughout the journey of like an hour about we talk about constructivism, collaborative, integrative, reflective, inquiry based learning, and all of those are valid in terms of learning, there is no one true way of doing it. And I think all of you get your needs to find out like, okay, for that application, I think maybe security works for that application carrying security work, and also ozotic, the machine of course, out there, there is multiple of them. So I just want to share one thing here about how we build on learning theory to create a layer of learning, which is not a single theory alone, that could ever be accomplished. So I would like to start by saying, the structure of feedback and credit mechanism act in critic as your enforcement for learning behavior. So the calibration activity, the grading power adjustment, the feedback mechanism, it can have like a reward, because there's a gamification there where you know, you want to gain points, so on, which is based on theory of behaviorism. And then there's another layer on top of that, which is like the feedback mechanism to extract higher order learning like critical thinking skills, problem solving skills, and all other skills, which is related to like the connectivity cognitivism, where we want to enhance the recognition of the students. And then now more wise, I have been searching for a long time, but I think AI is not proven yet. But does AI enhance cognition or not? And I haven't find the answer yet. So I still need to find you another debate as well on that. But for now, I left it as is, is like we, of course know, it's proven that students among others want to enhance their knowledge based on critical thinking skills, they enhance their cognition towards a problem, because you don't give them the answer. They go through the journey.

Gabi Immelman:

And I was just wanting to check are you think cognition cognitivism,

Carine Marette:

the cognitivism theory? Yeah, so I just talked about behaviorism. Now just talk about cognitivism. And then on top of that, there's two more layers, which is the layer of cognitivism that where the feedback mechanism and critic app pair to construct knowledge, the constructivism process help to further construction of knowledge among each others. And then it's a well rounded a holistic approach to learning that has been proven in the literature, because there's no new concept. And then the last one, of course, is like to, you know, we talk together about it as well, you know, the importance of social social approach. And then on that one, I want to talk about some theory of like Bandura, Halo theory, critical pedagogy, and then began asking how we just like provide different viewpoints and scaffolding activity in order to learn with each others. And then this is kind of like in depth, access to viewpoint these days that we didn't have in the past and that transform the oppressive social structure. And I think this is what I would like to end this debate is like, education now is like a layered approaches in order to bring diversity into technology. And then without me or gayby to provide technology, I think it will be impossible for educators to reach the potential that they want to so I encourage, I guess, own educator to learn about the technology, and then just find out what's the best fit for them. Amazing.

Alexander Sarlin:

I know we're coming on time. So I want to give each of you a moment to you know, just give a little bit of a closing statement about your approach to pedagogy. I think just one quick point before that, you know, your phrase earlier, Gaby about the cost of intelligence is going down, really stuck with me. And I feel so relevant to both of your approaches, right? It obviously is relevant to AI in that the cost of intelligence virtually created intelligence has gone down so much that you can introduce it in all sorts of frameworks and contexts and technologies. I think it's also relevant to peer assessment and peer learning, because the ability to communicate on the kind of scale the asynchronous communication, synchronous communication, mobile communication means that access to other human intelligence has also gone down and that you see that in things like Reddit and Stack Overflow, and all sorts of environments like that. So I think that is just something that stood out to me as a really interesting shared approach. You obviously share a lot of ideas here. So let's go into our final statements. Gabby, let me start with you. We're almost at a time just in a minute or so. Tell us about your sort of total thinking about why AI tutors provide a really strong pedagogy for learners.

Gabi Immelman:

So like, if you've embedded on like, just to go Are two thoughts that Karen was sharing and something we care about is constructivism. So puppets work by God's keep and Dara social learning theory. And I also really care a lot about, you know, Ryan dassies work on self determination. I think that for me is one of the things we care most about is how do you help young people maintain their sense of motivation in a world where human and machine output is indistinguishable? And I think that's something we are only just starting to grapple with, like, what is that gonna do to the psyche of humans? How do we build motivation? How do we find connection? How do we stay motivated to build mastery and skill and pursue meaningful experiences and meaning making? How do we practice agency? So in terms of the pedagogy that underpins what we're trying to engender and build and cultivate is fundamentally to support young people like at the end of the day, our end users, young people, and we see teachers as a fundamental ally, in supporting developing young people's sense of motivation, and their desire to want to learn in the same way as they want to use tick tock. And so I think, you know, one of the questions we ask ourselves is, how do you engineer communities that value learning, and that learning is human endeavor. And maybe learning is also a machine endeavor, but expression of creativity is a really interesting one to explore. So that's a whole nother debate is what is human creativity versus machine creativity. But I think we often confuse the idea of creativity and self expression. And so I think what's really important as a society is how do we wield tools like AI, to help us express ourselves and find connection and solve problems, but not to confuse that with this idea of creativity, which is the pursuit of intelligence that's trying to come up with novel ways of solving problems? Because I think we might trip ourselves up. So yeah, those of you my closing thoughts, we want to just create communities that learn together. And that's the social experience.

Carine Marette:

Yeah, I couldn't agree more with like a be saying, and then for to support her argument, I want to say, a quote from Ken Robinson, which I really love this person I racing piece for him for what he had accomplished for education. He said, Curiosity is an engine of achievement. Wishes goes to now my final thought to answer your question, Alex is Kenick is about like asking people, What do you want for your students post graduation, and I think is about ultimately is to make learning, engaging, effective and efficient during the courses of like, during your course. And then make that as much as possible real to the work experience. For example, right now, I'm doing my PhD in education. And one of my assignment is, hey, debate with your classmates. While I said okay, I'm not going to debate with someone as a fake debate, why not doing some intelligent conversation with Gabby, and then I had so much fun today, then and now we can just say, Hey, this is uh, Alex giving a live podcast for the audience. And, and, and people are learning from it. So ultimately, I will say that final statement is, let's make sure that our students get smarter, post graduation and use as much as possible they have time to build a portfolio to display to showcase because we want them to get the job, we want them to just like really say, hey, in that university, I build this kind of like, learning and and this is what I can display. And I recently just as it was really straight. Two weeks ago, I went to do ust they asked me, okay, do you want to volunteer to present? I say, why not? So I present what Kitty is about. It's a class of engineer. And then they have to basically do a project on how to enhance design and you know, part of the Moodle platform, and I also share the history and now they are working on it right now live. And then they have to present to me in two, three weeks. And then when they will present to me, I told them, the winner will get a letter from us for being like do awesome job. So it's kind of like giving, like the great motivation as you really emphasize KB and then I will do it. Like I really want to help them to just like get a job and happy to participate in more universities classroom as well. Yeah. Fascinating,

Alexander Sarlin:

really interesting debate about two incredibly important pedagogical approaches, AI tutoring and Peer Assessment. I really appreciate you both being here. And you know, listeners may notice there were a lot of theories learning theories dropped during this debate, we heard about behaviorism cognitive vism constructivism Rhian determining, well, yeah, all your social, cultural, amazing stuff constructivism. So we will put a set of links in the show notes for this episode of, you know, definitions of some of these theories as well as some papers you might want to look into if you want to dive deeper into any of these pedagogical theories that these two amazing founders are using in their work. I want to thank both of you Carine Marette from Kritik and Gabi Immelman from Mindjoy. Thanks so much for being here with us on edtech insiders. Thanks, Alex.

Carine Marette:

Thank you so much.

Alexander Sarlin:

Thanks for listening to this episode of edtech insiders. If you liked the podcast, remember to rate it and share it with others in the tech community. For those who want even more Ed Tech Insider, subscribe to the free ed tech insiders newsletter on substack.