Transforming Education: Understanding AI’s Role in Student Engagement

Hello everybody and welcome to the TeacherCast educational network. name is Jeff Bradbury. Thank you so much for joining us today and making TeacherCast your home for professional development. On today's show, we're gonna be having a conversation with two co-founders of a fantastic AI company called Kira.

And we're gonna learn not only what makes an AI company tick, but how you and your school district can create meaningful professional learning to not only go into the applications, but through them at light speed this year. Look, let's face it, we all know that AI is here and it is in our classrooms. It's just a matter of learning how to use it, how to leverage it, and how to use it as a tool in our toolbox to save time. So stick with me today for this entire episode. This is one you are not going to want to miss.

course of this is the first time you're listening to our show. Thank you guys for joining us. You can find all of the episodes over at teachercast.net forward slash podcast. And if you'd like to be featured on the show, head on over to teachercast.net slash contact today. We would love to feature you and your classroom on our next episode. My guests today are both the co-founders of a fantastic educational company called Kira, the first AI platform that delivers personalized education at scale. is an absolute fantastic program. And I am pleased to bring on

both of the co-founders Andrea Pasenetti and Jigridi Agarwal. Andrea, how are you today? Welcome to TeacherCast.

I'm great, thanks for having us.

Thank you so much for being here today. I'm looking forward to having this conversation. Before we go much further, tell us a little bit about Kira and what can we expect from Kira this year as we head into school?

Jarki and I started Kira a few years ago. We actually started Kira as an AI company before it was popular and cool to be an AI company. We started the company with a very specific vision and mission at the time, which was to empower all instructors, even if they weren't subject matter experts, to be effective instructors in whatever subject they were teaching. This was at the time when...

flipped classroom models were getting a lot of traction. There were a lot of teachers providing instruction and subjects that maybe they hadn't studied themselves. They hadn't had formal exposure to academically. But by dint of being in classroom settings and environments where flipped classrooms and the ability to deliver lectures over video was increasingly democratized, put them in a position where they needed to answer questions from students that

were more in-depth than perhaps their backgrounds allowed them to answer confidently and accurately. And we started building this as an AI teaching assistant. Ultimately, technology has evolved pretty substantially. The frontier of AI has evolved very substantially. And so we've been able to take that original vision and expand it to be a full suite of AI tools for teachers, administrators, and students.

designed to make personalized learning possible at scale. And that's what we're here to talk about today.

How did the two of you meet and how did the company form?

So we started the company with Dr. Andrew Ang, who is a professor at Stanford and also the founder of Coursera. We were both students at Stanford at the time. I had actually recently graduated. Jharati was still in a graduate program at school. we met just by dint of circulating information about this vision.

You know, this is the school year where it's without a doubt AI is here. AI is now in our classrooms. think last year was kind of the, is this? How is it going to work? I'm kind of nervous about it, but even throughout the summertime as AI driven ed tech companies were making the announcements that nope, this is in your applications. We are building AI into everything. I think there's still a lot that teachers need to hear answers from.

from their ed tech companies. So let me throw a couple of topics at you that I know are gonna be on the top of everybody's mind. The first one of course has to be student safety. Talk to us a little bit about what you guys are doing in your platform to make sure that our teachers and our students are safe and our personal property stays where it needs to.

I think I'll tackle it in two parts. One, I mean, because I think safety can mean many things. think one, there's the data privacy, data security aspect of it. And then there's also safety in the sense that when a student is using an AI tutor, if they reveal information that is concerning or maybe about self-harm or something like that, how do we make sure that the right people, their teachers, potentially parents, how do we make sure that that information is also relayed to the right people?

So I'll start with the data security and privacy side. So we are very careful to not share any PII ever to like any models that we use. So primarily right now we're using various open AI models. We're experimenting with different ones, whether it's open AI or cloud, or we're seeing which ones kind of work best, but any data. So let's say in our AI tool.

Any data that is sent to those models is anonymized. And if you look at those companies, they also have in their kind of privacy documents that there's a setting essentially that you can choose to make sure that they won't use any of that data to train their models, so especially for enterprises. So that's kind of taken care of from that end. And then on the other piece around, I give a student reveals information that

could be highly concerning around their safety. We actually have a system in our platform that will immediately notify their teacher. Because our platform has both a student interface and a teacher interface, right? So as a student is taking a course, these AI tutor, they express something concerning. The teacher will immediately get notified that the student has said something and the AI tutor will also.

be like, you sorry, I can't respond to this and notify the teacher. And we also let the students know that the teachers always have full access to their entire chat history as well.

I think that's important that not only do teachers understand the platform, but also students and parents. I'm curious when you're working in education and you have all these different communities that you have to serve with your tool, what has the feedback been like over the last few years as the company has grown up a little bit? What are teachers saying about the platform? What are students saying? And have you had a chance to work with many parents?

Fundamentally, the backdrop that we operate against has changed a lot since we started Kira. Attitudes towards AI had really shifted. Two years ago, there was a lot of apprehension. And in fact, I would say fairly strong resistance to adoption of AI, especially in classrooms and especially as it pertained to students. The number of options available at the time was also limited.

And interest in learning about AI was also fairly low. And there was also resistance on that front as well. Whether it was for ethical reasons or in some cases political reasons, there were a number of barriers. And as we all know and doesn't require much for laboring,

the entire landscape has shifted in the past 18 months. And a lot of that had to do with the introduction of chat GPT and the commercial availability and also consumer availability of LLIMES. So I think the environment that we operate in and the readiness of audiences to engage in the conversation has really shifted. I think there's also been a renewed level of enthusiasm that's come with that shift as well as curiousness

interest and desire to leverage the technology to improve outcomes across the board, whether it's academic outcomes for students, completion rates for students, or increasing the efficacy of instruction or making teaching a more sustainable undertaking. So the feedback we've gotten has been extremely positive. One of our biggest rollouts, so our first big test case as a company

was assisting the state of Tennessee with the rollout of their computer science requirement. So in 2022, Tennessee passed legislation making computer science a requirement for graduation for all high school students. This means a state where a small fraction of students had historically taken computer science was now ruling out a requirement across all of its high schools or by every high school student needs to take one full year of computer science in order to be able to get a high school diploma.

which is a big undertaking for a variety of reasons that are probably fairly obvious. One of them is there aren't that many teachers who have backgrounds in computer science. So there are a lot of first time computer science teachers and the need to ramp them up on a very accelerated timeline so that they were ready to welcome students and be effective instructors in little more than a year was a pressing need. The second was supporting students who

would be taking a fairly demanding subject and course in the event that they weren't able to get answers or immediate support from teachers who were also backing the subject for the very first time. And so the tools we rolled out, an AI teaching assistant and an AI tutor for that particular use case were two important ingredients to the successful rollout of that.

requirements and law in the state of Tennessee. This was fairly early on in the AI journey that schools and districts are on, and it was a very natural use case and one that was informed by an immediate and very pressing need. And the feedback we got from that was extremely positive and also helped inform our roadmap beyond computer science and into other AI tools as well.

When you're working with teachers specifically, what kinds of things are you getting feedback on that they're looking for in their tools? Specifically something like Akira, where you've got a toolbox worth of different AI tools at your disposal. What are some of the things that they're asking of you when they're giving feedback?

You know, teachers are incredibly creative and resourceful. And the wonderful thing about working with teachers in the realm of AI is that they always have new ideas for ways in which the technology can be leveraged to improve student outcomes, to make learning more interactive, to make learning more engaging for students, to make teaching a more sustainable endeavor. So

the amount of product feedback and engagement we get is enormous. And one of our guiding principles, and I think what AI enables technology companies like Pira to do is basically remove any boundaries for what's possible or make the boundaries very, very far out. When we engage with teachers or administrators, they often say,

it would be great if we could also do this or this is awesome. Can we then go ahead and do this other thing? And we want the answer to that question to almost always be yes. And if it's not yes immediately, we want it to be yes at some point. So I think that's a really exciting part of this work, both as a function of how teachers operate and how industrious and

creative they are and what the technology allows vendors like Kira to do in response to that creativity and sort of thinking that pushes the boundary.

You know, when a school district is looking to bring in an AI solution, much like here, that's they're going to be used for teachers to create, you know, dynamic lessons and activities and a whole mess of things. One of the things that they always have to ask is, is this going to make an outcome in our classroom? Is this going to be something that is positive?

for our students. Obviously right now budgets are tight and there's so many different applications. I can't think of a single school district right now that isn't looking at their spreadsheets of applications and basically trying to slash them into two at least half if not more. Where does Cura fit into this? Talk to us maybe about, you you don't have to give me specifics, but a little bit about the financials of a student if a school district's looking to bring Cura in. how do you see yourself making an impact

in the classroom at the end of the day.

This kind of moment that we're in right now for education, it's new for the world, right? Like think we're in an extremely transformative moment for a lot of industries, education in particular, and it's a very high impact industry, meaning like, I very much believe, right, that the way that somebody's taught, what they've learned, it has a huge impact on the rest of their lives. And so...

I think that the truth is, that we're kind of all figuring it out and it has to be like the way we develop our product too. It has to be very iterative, right? So the approach we take isn't one where it's like we build something and we definitely say this is the right approach immediately, right? Like we, talk to teachers as we're designing, as we're building, because as Andrea was saying, they have creative ideas. They know what it's like to be in the classroom and

You can never replace, like, you know, if a teacher tells a student, like, I believe in you and I believe in your potential, like, I don't, you can't replace that, right? So that's just, I'm kind of saying that comment to lay the groundwork for my next comment around student outcomes, right? I think there's a couple of ways in which we're doing this. So one thing that we're going to be rolling out fairly soon is actually,

So our platform will give teachers kind of an understanding of how students are performing in various, for various standards or various skills. So what are the things that students need support with and what are the things that students need, that students are good at and can potentially even help other students with, right? Like that is the ultimate, like, I mean, doing that at scale is the ultimate challenge for teachers, right? And so, and I think with AI, we truly have an opportunity.

to do that. I think, of course, doing that accurately, takes a lot of data and it takes some time, right? Like we need to have enough data about, it could be about student submissions, it could be about students interactions with the AI tutor, it could be about like a bunch of different kinds of data in order to make that judgment. And we want it to be an informed judgment. I think a big part of this too is actually around transparency. I think we teachers,

Teachers need to understand, and this is our responsibility as well, to make sure that they understand why the AI tools are outputting the judgments that they are. So if some tool says, hey, I think a student is good at this subject or good at this skill, why? Why is it saying that? And so a big part of what we do too is actually around AI education and AI literacy as well for educators, for teachers. But then within our platform too, it should be really clear.

why it's coming up with the claims that it is.

I think one of the things that teachers are going to run into these days, specifically this year that we're coming up to, and I know I'm feeling this too. I've heard a lot of conversations, specifically around, you know, when we were at the SD conference, there's not a lot of well put together professional development yet. There's a lot of PD that is, this is AI, this is what we're doing. And specifically, I was talking to a few other AI companies recently and like,

The concept goes, we're doing a lot of 101 PD. How do you use? How do you prompt? What does this look like? How not to be afraid of? But I haven't seen a lot of 201, 301 training. What does it mean? What is the difference between this prompt and this prompt? So let's talk a little bit about this here. If we can take us a little bit more in depth for teachers who are listening to this. How do we really make the most of Kira?

or other AI tools that we have, it's different, right? You hear this in different commercials now, you know, they're using AI as a search engine, but when you use it a different way, then it can be so much more powerful. What suggestions do you have for teachers who are really looking to dive into Kira or any AI tool at this point to really get beyond that basic search compatibility and really make it be a tool that they can't.

go without every single day. How do you dive deeper into these conversations with the tool?

I think as with any big paradigm shift, there's a lot of noise. There are a lot of people who either have informed insights or uninformed insights, but a lot of people who want to talk and share their perspectives and end up filling airwaves with just a lot of sound and noise. And I think cutting through that is very difficult and it can feel very overwhelming.

I would frame it and maybe a simple framework I would suggest is to partition learning about AI into two categories. There's learning how to use AI tools. What can AI do for you very practically? And what do you need to be aware of when you're using AI as a tool or a set of tools to do those things? And the second category is how does it work? Like what is actually happening under the hood to make those tools

produce the output that they produce or render the outcomes that they render. The latter is a lot more challenging. The latter is often a more academic undertaking. And there's really no end to the depth that you can pursue the how question. So I would suggest approaching the how.

if it's a source of interesting curiosity, but to not be overwhelmed or not feel a huge amount of pressure to understand exactly how AI is rendering the things that it renders. I think in the former category, namely, what are the tools, how are they being used, and what is important to know in the way of privacy and data security, in the way of bias, in the way of how those tools operate,

I think that falls on the tool providers. And I think as with any application, conceptual learning can only go so far. I think the best way to learn the how of tools is to use the tools and lean on the providers of those tools to make the experience of using them intuitive and informative by design. So one example is

Kira has an AI grading tool. When we auto grade a submission by a student and share that submission with a teacher, we provide insight into why AI graded the assignment the way AI did and provide reasoning around the outcome that was generated. We also require teachers to review those grades before they're released to students. So I think there is a design problem.

or there's a design challenge and a design responsibility for the creators of those tools to make the utilization of those tools informative and educational without creating unnecessary friction. One thing I'll add is I think AI tools in education are undergoing kind of a generational transformation. Last full year was the first generation of AI tools being adopted a little bit more

universally in schools, or at least on a broad base. We call that generation of tools, prompt wrappers, rather than GPT wrappers. And by that we mean they're a way to help first time users of chat-based interfaces write prompts that are structured in a way that generates a good streamed output from an LLM. So if you're creating a lesson plan or if you're creating

an IEP, for instance, you can go into a chat GPT or a plot on the assumption that you have an account where the LLM is not being trained with the data or you're removing sort of identifying data from ear input, but you can create a prompt fairly easily that will generate what it is you're looking to generate. And what the first generation of prompt wrappers did was just make that process even easier.

they would take, you know, they would give and they still do give users guidance on what kinds of inputs or what kind of information to include when trying to generate an IEP. I think by and large, those tools have served their purpose. A lot of the teachers we talked to, a lot of the administrators we talked to are ready for tools that deliver more value. They're ready to see things that do end to end work. They're ready to see things that maybe

go further than providing texts and actually generate things that students can interact with and then provide insight into those interactions themselves. I think that's what's gonna happen this year. I think what we're gonna see is a much broader adoption of much more powerful tools, which puts the onus on the providers of those tools, again, to make them informative while they're being useful. I think getting teachers together in a room of, you know,

50, 100 teachers and providing a PowerPoint based professional development around AI is definitely not unhelpful. But I don't think it provides the foundational knowledge on an ongoing basis that will make teachers and administrators responsible users of AI as that AI technology evolves and changes and becomes more powerful. So I think it really comes down to that.

We have a lot of instructional coaches that listen to this show and being of that ilk myself, we're used to exactly what you just said. We have a slide deck, we teach, they do, we teach, they do. And I think there's a place for that. Then you start doing teaching on search engines and it's easy to say, find, you know, go into a Chrome type in, the, the name of this company and it happens to be the first one. And for the most part,

Everyone's going to get a similar experience. But even for the three of us on this podcast, if I said, open up chat GPT and type in this sentence prompt, we're all going to get different things. And as the trainer, that just makes my job even harder because now I have three Italy in this podcast, three different things that I have to look at troubleshoot, ask questions to, and you can expand on that with everything and

when you're dealing with education and teachers, you've got the entire bell curve of learning and acceptance on AI. What advice do you have or even with your own team trying to train other school districts on your application? What are some of the best ways that you've seen school districts adopt a platform district wide school wide building wide?

I think one thing I like to tell teachers and educators is that go off the assumption that the first output you get from an AI tool is not going to be exactly what you want. I think what often happens is like it can feel really cool. It can feel like magic when you get some sort of output, whether it's a lesson or a grade or something.

That's very powerful, powerful output in a matter of seconds, right? But it's not magic. Underneath the hood, there's a lot going on. assume that it's not... So if they assume that it's not going to be exactly what they want, then they will need to iterate on it. And I think that's where the training comes in too, like training people how to iterate on output from AI and building in that intuition of like...

how it actually works, where it comes from is so critical because we have a set of AI tools now, AI products now. I don't know what the world is gonna look like in a year, in a month. So the applications, the AI applications are probably gonna look vastly different a year from now. So it's not, don't think it's enough to be comfortable using the AI tools that are available today. They need to build an intuition about what's kind of going on underneath the hood. And they need to know that like,

Almost always you'll have to iterate on the output at least a little bit.

You know, speaking as a middle school teacher, we don't teach conversational skills very well to our students. We don't teach small talk versus business talk versus like personal talk. And I feel that as I'm going through my journey of learning AI and getting to know all the different tools, that's a skill that I wish I had a little bit better because you know, walking into a Gemini and going, Hey, I need this like that's 101, but really create this is another level, but you are a, and then.

synthesize these things. Those are the conversations that I'm looking forward to having this year. Those are the conversations I'm looking forward to working with my teachers on. mean, even as we're sitting here having this recording, it's the middle of August, and I'm looking at my email, and I just had four of my favorite AI applications send out an email that says, Hey, GPT five just dropped. And here's all of our new features. I'm going, well, since we've started this podcast, I no longer know how to use the applications that I've been using every day.

My world just changed in the last hour. I see you guys are smiling going. It's it's never going to end like and what was it Sam Altman's quote or something like this is the dumbest that that open AI is ever going to be or whatever that quote is because it's just going to keep getting better. I think having a solid foundation in your school for PD having a solid foundation of trust teachers are going to make mistakes. Students are going to make mistakes.

coaches are going to make miss ed tech companies are going to make mistakes. How do you move from there? How do you how do you support all that? guys, I want to say thank you so much for coming on the show today. I you as I've say at the end of all of my shows, the microphone is always on. I would love to continue this conversation, but, first I'll go to geography and then Andrea, I'll give you guys the last word here, but as we're going into the new school year,

What do you want teachers to know and be thinking when it comes to working with their AI tools? And then tell me one thing about your application that you'd love to have somebody who's brand new check out first.

This is not an easy time for teachers. mean, teachers have always had a very, very difficult job and heroic to say the least. But I think this time especially is challenging to another level. Teachers have to be learners in a way that it's just, it's hard. It's really hard. And I think one thing I'll say is that, just I'll remind teachers to not forget.

the impact that they have on students just with their words. Like I said this earlier in the podcast, right? Like I still remember my 10th grade English teacher who, like I was not a good writer for a while, for a long time, right? That teacher spent a lot of time with me and she said she believed in me and all those things, right? Like I will never forget that. And so I'll tell teachers, don't just, all these AI products, what we're doing, it's all with the goal of helping teachers, the impact that they have.

on their students is, yeah, mean, hopefully it'll only grow and it just, should not forget that. I think I'll say that. What was the other?

It's one feature. I'm a brand new user. First thing that I said, check out.

yeah, well, that's a tough one because we have so much. But I will say one thing I'm really excited about that we're rolling out pretty soon is our lesson creation tool. And this tool will allow teachers to, it uses AI in the back end and they can create an interactive lesson. this lesson has activities that students can actually take lots of different activities types and they can create the initial version in a matter of like minutes pretty much. And then they can keep iterating on it to get.

they want and personalize it to various student needs. But I'm really excited about that tool.

Andrea.

I would say that from a district perspective.

When you're buying an AI products, you're not really buying stockware as a product. I think you're buying a partnership. You're buying a service to your point. The best AI providers, the best AI tools, the best products are going to be very evolutionary. And I think districts should be looking for.

for providers and vendors that will be partners in that journey rather than static purveyors of a single product. And I think that's a very different way of interacting with software providers in education, moving away from point solutions or single purpose solutions to software and AI truly as a service in a way that evolves as a technology evolves in a way that can

help demystify some of the questions you shared just a few moments ago when OpenAI drops a new model, for instance, what does that imply for your use case? And a partner that is both knowledgeable, able, and available to help navigate those situations and questions is what I think districts should be looking for. this is something I say in a way that's very earnest and not in any way beneficial for companies like Kira.

I would avoid multi-year partnerships at all costs. think the best providers today should be and are able to.

basically engage in one year pilots. And I think districts should work with multiple partners to the extent that they have the bandwidth to interact with them on one year pilots to see which partnership feels the best for them in terms of getting really meaningful value for students and for teachers. Obviously there's an upper limit in terms of times and best and bandwidth, but it is an opportunity to really put vendors to the test.

and make sure that schools are engaging on this really important and inconsequential journey with a partner that they're competent in and is suited to working with them. In terms of what we're ruling out and we're really excited about, lesson generation, I think is going to be very exciting for teachers and administrators as Jarrity was just saying. We really believe that

AI should help generate high quality instructional materials. Schools spend so much money on curriculum and that curriculum is static. It is one size fits all. You buy a textbook, it is not going to change. The words on the page are not going to become reordered. But everyone knows that learning fundamentally needs to be very personalized.

not everyone is going respond to the same textbook in the same way. In fact, everyone is going to respond differently. And we believe in a future and in fact, an immediate future where personalized learning is enabled by personalized instructional materials. We think every student should receive content, instructional content and curriculum in a way that's tailored to that student's learning needs and learning style and an approach to learning and where assessment

follows the same path. And we think that's going to be what ultimately results in asymmetric learning outcomes for students and districts. So we're really excited about that. It's a big launch for us. And we welcome everyone to try it for free on Kiro's website.

Well, I invite everybody to check out here. I've been playing with it now for the last couple of weeks heading into my school year and I absolutely love it. Easy to use quick as anything. Andrea, where do we go to find out more information and sign up for Kira?

www.kira-learning.com

And we're gonna make sure that we have the links to this and everything that we've been discussing so far over in our show notes over on teachercast.net forward slash podcast. Andrea Dragwetti, thank you so much for coming on the show. I really appreciate having you guys on. Let's call this part one. Looking forward to continuing the conversation.

Thanks so much. See you on part two.

of course, as we head into the school year, there's going to be a lot of questions that you guys have as far as how to use AI, what types of AI tools keep your bookmarks over on teacher cast that now we're going to be going through all of our favorite applications and going through them one by one to show you guys not only what but how you can use them. If you're an instructional coach, don't forget to head on over to your digital learning coach dot com or all of our instructional coaching resources are. And that wraps up this episode of digital learning today on behalf of everybody here on Teacher Cast. My name is Jeff Bradbury, reminding you guys to keep

Keep up the great work in your classrooms and continue sharing your passions with your students.