AI Literacy Series Ep. 16: Karim Meghji, President and CEO of Code AI, on Empowering Students and Teachers and Developing Human Agency in the Age of AI

Speaker A: Hi, I'm, uh, Miriam Vogel with Equal AI, and we are so pleased to have you join us for this edition of In AI We Trust. Welcome back to In AI We Trust. I'm your host, Miriam Vogel. And today I am thrilled to introduce you to our guest, Karim Medji, the CEO of Code AI, a global education nonprofit on a mission to ensure every student has the opportunity to learn computer science and AI. Code.org started back in 2013 as a small, volunteer driven effort to bring computer science into the classrooms, and the demand has been overwhelming. It has grown quickly. Today, code AI has reached over 100 million students, 3 million teachers, and partners spanning the globe. Karim joined the organization in 2022 as chief product Officer, where he helped lead its evolution towards an AI centered strategy before stepping into the CEO role. That expansion has made real impact. More than 25 million students have built and participated in Code AI's Hour of AI campaign. 6 million have built foundational AI skills through their core curriculum. And this work could not be more timely. We are so excited to talk to you, Karim, about how you meet this moment with curiosity and purpose and hope. So let's dive in. Karim, welcome to the show.

Speaker B: Thank you, Miriam. It's great to chat with you today. I'm excited about the conversation we're going to have and, yeah, all the good things we've been doing and all the things we hope to do.

Speaker A: So you're giving us a lot to talk about, Karim. But before we dive into the current mission of your work and upcoming announcements that we'll get to make on the show today, I'd love to introduce our listeners to your journey to this space and an understanding of, um, how you got to be so excited about the technology you're now leading. You've described a light bulb moment in high school where a teacher helped you break through a wall and how that fundamentally changed your approach to problem solving. Can you take us back to that moment? What happened and how did it set you on the path where you are today?

Speaker B: Yeah, it really was a sort of, uh, a, you know, one of those moments of life where you wonder, what would the other path have looked like? I was in a math class in high school doing calculus. Sometimes I joke. I was sort of a lazy mathematician. Didn't love doing the longhanded calculations required to, to actually solve the calculus problems that we were posed with. And one day my teacher said, hey, you know, if anyone's interested in seeing how computers can help us solve math problems, come after school and we'll work with you. Know some basic programming and we'll learn how to do that. So just within a few hours, a few of us that joined him after school had built programs on the, you know, one of the earliest Apple computers to actually calculate these math problems. That would've taken us 15, 20 minutes by hand. And here we were doing these problems in seconds. And there were two things that, that really did for me. One, it. It gave me this sense of agency, this sense of empowerment, this ability to, to direct things in the world, not just be directed by them. In this case, a math problem was directing me. It owned me, but now I owned it. So that was one really interesting thing. And it sort of sparked this curiosity about, you know, technology generally speaking. And so from that moment, um, forward, I spent time pursuing a computer science education path. Went into software engineering, building technical products for the last 30 years, and then was really attracted to the Code.org work. When I joined in 2022, we were still Code.org, because the thing that struck me was I happened to be in the right place at the right time, had the right person, everything was right for me. And code.org has been on a mission to remove the right and get every student the opportunity to learn about technology, to find their agency, and, um, hopefully sparked that curiosity about what technology can do for them.

Speaker A: I love that someone who feels agency and focuses on it is leading in this effort, because it's what I think is one of our greatest concerns at this moment, where very few people feel agency. I'm sure you find this in your conversations too, whether it's CIOs, CEOs, boards, consumers, general public, politicians. Most rooms I walk into are not feeling agency in this AI conversation, which is exactly the opposite of where we need to be as we make these pivotal decisions at this consequential moment. But you've been doing this a while, so there's a lot to learn from the organization and from your focus. As we talked about your organization then, code.org launched 13 years ago. Now that's wild. With a focused mission to get computer science into the classrooms. So here we are over a decade later, and you've reached over 100 million students. Kudos to you. That is so significant and should be really celebrated and acknowledged. And now you've talked about this new significant pivot to code AI, putting artificial intelligence at the center of what you're doing. Can you talk us through that change? Was it based on an internal debate about how much to lean into AI? And how did you decide that AI should Not just be part of your curriculum, but the curriculum.

Speaker B: Yeah, yeah. First Miriam, I want to start by saying, you know, the 13 years that we've spent doing this work, it just wouldn't be possible without a host of other partners who really make this work real. Our ability to reach students is, it's pivotal to have teachers who understand the concepts that we're teaching and engage those students. So we wouldn't reach the students we do without teachers. I just want to start with that, like the teacher community. Kudos to them for being on this journey with us, going on this journey with us to educate our students. And then there's an entire community around the technical sciences that we teach today. A computer science community, a developing AI science community, a data science community, and we partner with folks from across those communities to do this work. So it really does take a village in today's day and age to make this sort of system change real and to actually reach the level that we have in terms of the shift in pivot. This is one of those pivots that's actually been in motion for multiple years. I was just looking at like the clock, reminding myself of what time and date it is. Because if I think back, code.org in 2018 started to teach about AI. And um, the reason is because AI is probably, at least today computer science most innovative, most interesting, most exciting innovation. It's probably the thing that, you know, when we look back, will be one of the most seminal moments in computer science's journey, if you will. But at the same time AI is changing computer science. So there's a really interesting dynamic occurring here. And I raised that because AI has been a, uh, part of what we've, we've taught about. I think what's happening now is we taught about this much computer science and this much data science and this much AI science. And what's happening is these things are starting to balance out. What I sometimes will call the digital sciences are, ah, the focus areas that we as an organization have. And that's the recognition we've had over the last two to three years as increasingly we're seeing acceleration, velocity. Some of why you described those CIO rooms that you spent time in where there's a lot of like, lack of agency. The speed and pace at which this technology is changing leads to that. And I think we believe that education is the way to sort of rein that back in, give agency back to our students, our humans and, and understand and give them the understanding about how these tools and technologies actually work so they can do more with it versus feeling like it's happening to them, so to speak.

Speaker A: Yeah, well, um, all important at this moment, and I'm so glad you have found the way to engage so many. So let's just talk about one of your flagship programs. Over 25 million students, I understand, have participated in the Hour of AI and that's a significant number. That's a number of people that you've touched and impacted their day, their life. It sounds like some of the offerings that they have participated in have been just delightful. Generating music sequences, programming droids from the Star wars universe. I'm, um, curious about the philosophy behind these experiences. What have they taught you about what sparks a young person's curiosity? And can you tell us about what some of these projects look like?

Speaker B: Yeah, yeah. I love talking about the Hour of. What was the Hour of Code now is Hour of AI, Miriam. So I'm glad we're going to spend some time on this. We're actually over 33 million learners now who spent an hour with these activities. And it's not just our activities. There's an entire community who contributes activities and participates in the Hour of AI. It's truly a global learning event. But again, our, uh, community is what it takes to create this change. These activities are hour long. They're built for, you know, a child that is five years old, up to 105. That's how we think about it, uh, making them fun, engaging, but really focusing on a learning outcome and learning objectives, how we approach our work and how many of our partners approach the work. The intention is if you can engage a young person and sort of get them curious about the why, the next thing, that's how you pull them through from just one hour to 10 hours to 50 hours of deepening education and learning about technology. So how we approach the work itself is we try to make the activities fun and very age appropriate. Our younger students can have dancers dancing to music, and it's music that's popular music that they would know. And our older students can work with a, uh, music platform where they get to create their own music or again, use mixes from popular music. So, you know, we try and think about what is the, the way to engage students, given where they're at in their sort of progression and growing up, if you will. And then we wrap that nice packaging around a core learning objective, something that they can walk away with saying, oh, I learned something new. And our most recent activity, the one that Code AI produced for this last Hour of AI mix and move with AI we did the combination. We had students prompting the design of a character that would become a dancer and then we had them prompting the creation of music that they can then adjust with programming and putting those two things together. So I've created a dancer, I've created some music, I put them together, the dancer moves to the music. It's fun, it's engaging. They get to learn something, they get to play with some popular music and they have something that they can show mom, dad, the teacher, their, their friends, which is equally important. Right. The ability to exhibit that artifact after the fact is a great reinforcing confidence building and hopefully ultimately taking the next step, learning more.

Speaker A: And you mentioned that you've been able to be successful in taking them from the hour to the 10 hours and so forth. What can we learn from that process? You talked about making it fun, which is critical. Are there other threads that we can all learn from as to how you turn that into long term engagement?

Speaker B: Yeah, yeah. I mean, the next step usually ends up being, you know, teacher led. In this particular case, the teacher might facilitate the environment, the experience. A lot of our activities occur in the school context, but they're lighter weight students, you know, essentially are doing the activity and are able to guide themselves through it. Right. So the next step is also engaging teachers to the point of how do we bring continuity into the equation? As much as we're engaging students, we're also building confidence in teachers when we do these hour of AI activities. Most teachers that are teaching these subjects today did not have in their undergraduate study. Computer science is a vertical that they went deep on or AI science or data science or some of these concepts that we talk about. So I think the journey is really a shared journey. Our beneficiary, our focus is on students, but if we don't bring the teachers along, if we don't give them the professional training, if we don't give them the support to actually understand how to teach the subjects that we are teaching these learning objectives. We won't take that next up to the five or 10 hours, even if the student is interested. Now, hopefully the student might go home, talk to mom and dad, and you know, there may be a follow on in that way. And of course our products are available for outside of the classroom context as well. But I do think a critical part is to make sure the teachers not only themselves are comfortable, but they see that spark in their students. That combination reduces the barrier for them saying, I can do this in class this fall or this spring and I'm going To bring in this curriculum to do that, Whether it's code AIs or someone else's, that's irrelevant. But giving them the jumping off point I think is an important learning for us.

Speaker A: Mhm, mhm. Thank you for sharing that. I know another core commitment you have is accessibility. If these tools are not meaningfully available to different types of learners, there is a risk of leaving too many voices behind. So I'd love to understand and share with our listeners how that impacts your design process. And can you give us an example of how it's impacted some of the work that you do?

Speaker B: Yeah, yeah. So I'll interpret your question of accessibility very broadly and I'll start by talking not about our products. As a nonprofit, we, we do multiple things including policy work. And so our policy work is very specifically designed in many ways to ensure that students from all different backgrounds, household situations, with all the things that they have, uh, in their identity and who they are, that every student has access to the ability to learn about these digital sciences, computer science, AI science and data science. So it begins even before we touch any product, this idea of making sure that access is there, that accessibility is there in terms of the opportunity for a student to be either taught or elect to learn about these subjects. That's piece one, piece two. Are things like our AI getting students engaged, getting them excited, but it's also getting other stakeholders excited, it's getting parents excited, it's getting district leaders excited, and you know, not the teacher, but the educators that sit in the administrative part of the equation excited. And if you can get those folks excited, like the students are the bottoms up, the policy is top down, the teachers and administrators are the middle. So we're trying to actually activate all of those parts of the ecosystem because it does take all of those pieces to ultimately drive the outcome we want, which is an increased, not focus isn't the right word, but opportunity for students to really learn about the digital world, which is such a big part of the world they live in today. When, uh, it gets to the products themselves, we think a lot about ensuring that our products are built for different types of learners. So one very practical example I'll give you is block based coding. We use block based coding for younger learners. They, you know, uh, with dexterity, motor skills. As younger learners, it's easier to drag on an iPad or you know, even with a mouse blocks around the screen versus necessarily typing a lot. As they move through the grade levels, they'll get to text based coding and text experiences. But in our block Based experience. We realized that there were certain aspects of the block based experience that weren't as accessible on keyboards. So we spent time very specifically enhancing the library that we used. It's called Blockly. And uh, that's actually an open source library. A number of organizations use it. And so we spent some time enhancing that library and then enhancing our own products so that every one of those products actually used this. The new capabilities that we had added to the library to support accessible keyboard use, very important for learners that might not have this. The same typing dexterity or a visually impaired learner who might use the keyword actually to navigate things. And it has, you know, speech to text capability that help them navigate it. So those are the sorts of things that we think about to ensure that we can actually enable products that are accessible. The other thing we do, and I'll talk about accessibility in terms of access outside of the US So in the US there are certain things that are very, I'll call it, you know, pop culture that appeal to a young learner. But you take that same, that same curriculum and you put it in a rural town in India and it's not going to land. Well, the characters and the zeitgeist is very different there. So adapting to context localizing is another area that we spent a lot of time thinking about. How do we ensure that our curriculum is adaptable to different contexts. An example here is we have something called AI for Oceans. It's an AI based activity for young learners where they're essentially training an AI model on what is fish and what is trash and then they get to see it run through, evaluate its outputs and correct it as it goes. Well, we've just released a version of that for the India context called AI for farmers. So an agricultural society, rural, uh, society much more in tune with what that student is seeing around them. So that's another example of accessibility. So there's a few different angles that we think about, but they all come to the same point, which is meet the student where they are in terms of what they can do and what gets them excited.

Speaker A: Well, and I'm sure another top concern for you is one that everyone needs to be thinking about and that's privacy. And in your context, even more consequential with the student privacy. We know this is a primary issue for so many people now in legislation, litigation and strong headlines that we're seeing. And I'd, uh, love for people to learn from you how you are thinking about that balance between personalized learning, which often requires extensive data and protection of kids privacy.

Speaker B: Yeah, we're walking this line extremely carefully, Miriam. It's. It's such an important line to be on the right side of. We have, we're probably one of the organizations that's taken, I would say, you know, a very methodical and deliberate approach to bringing adaptive learning, or I'll call it tutoring, in this case just for shorthand, into the classroom context. We actually began working with AI tools on the teacher side first. The thought being support teachers in understanding how to guide learners that are at different places in their learning journey. Support teachers in actually delivering the curriculum well. So using that side of the equation as a way, an indirect way, to support students, because there's one person that knows each student well. Across those 15 to 20 students, there's one unique person that knows each of them well. It's the teacher. They're the ones who need the agency and empowerment to go and adapt the learning for the learners that are in front of them. Having said that, we do think there is value in bringing tutoring and adaptive learning into the classroom context. And so we're starting first, not with, I'll call it personalizing the way that a lot of people think about it, which is, you know, take in a bunch of data about a student and then base the adaptation on everything. You know, it has to be very carefully thought about. What we do is we look at, in the context of a particular lesson, how is that student interacting with that lesson in that moment. So their behavior right there affects how the AI will interact with them. So it's very localized. It does not include broader context. Right. So we want to start from the most narrowest circle and then layer rings in as we see value being added. And I think that's the most important point. Right. Instead of going straight to, oh, there's value here, widen the circle as broad as possible, bring all this data into the circle and assume it's going to work out, we're starting narrow, and then as we learn, take a step out. What more context is interesting, helpful, can guide a learner, Try it, test it, see the outcome, and then decide, is it, is it the right thing to do, like bringing that data in? If it is, and we can stay on the right side again of the student privacy line, then we do it. If it's not, then we don't do it. But having that sort of iterative approach, I think helps us find the right kind of expansion of adaptive learning without crossing a very important line that we do not want to cross. But giving that student and that teacher a capability that they wouldn't otherwise have.

Speaker A: And I'm hearing a theme from you today that I'm, um, hearing from a lot of thoughtful people in this field where you're only using the data necessary. You're looking at what is actually required and necessary, not trying to capture all the data and then making the decisions in a backwards way. And that sounds like what I'm hearing, a best practice among those who are really thoughtful about how to ensure that privacy is really protected. And I know another consideration for you is how to teach students how to think and not what to think. And that is really important right now. Not a lot of people have balanced that correctly. And it's something that's top of mind across our country and leading to a lot of polarization. And I'm curious to learn more about how you put this into effect in your curriculum. How do you think about this in terms of what you're offering and how do you apply this across the different age groups that you're thinking about? So what does critical AI thinking look like with a 10 year old versus a 17 year old?

Speaker B: Yeah, it's probably one of the most complex things to navigate. And there's a couple of reasons that, uh, I'll say that one is the technology is changing so fast. You know, if you look at even anthropic, just this week released the Fable 5 model. And I've been listening to some podcasts about the Bake off between even 4.8 and 5.0. And the change is dramatic in terms of reasoning capabilities. So understanding where the technology is and where it's going is a critical part of this question that you're asking. And because the puck is moving and we're just learning how to skate, it's a tricky balancing act. I'm, uh, just trying to be transparent. I think other practitioners hopefully recognize what I'm trying to express that we and I feel at this point. So I'll just say that as a backdrop, our approach has been to start with high school students with our work and really thinking about, given the step, the place they're at in life, given the step for them, we can kind of understand a little better what the world looks like now, or maybe even six or 12 months out, and try and guide our curriculum with respect to how models work today, how to evaluate the outputs, how to be a critical thinker, how to push back on a model, how to question a model or an AI system. And what we do is in our pedagogy, we guide them through that process. We actually create mistakes on purpose because at the end of the day some of these models are not deterministic. We can't assume it's going to make the mistake we want it to make. So our curriculum is structured at times to actually simulate mistakes and sort of like how do we teach a young uh, person to drive a car? We put them in a simulation before we put them behind a vehicle and let them drive down the road. Right. So I think about that approach as an important component. If you can give them a simulation of some of the failure cases, get them more sensitized to where, where to sort of draw that critical thinking line and how to push AI systems, I think that's a starting point to when they get into the real world, use real AI models, get into the workplace and start to hopefully apply some of those learnings in those contexts. And in addition to just evaluation of right and wrong, there are obviously things like evaluation of, um, bias. Bias can be very subtle. Right? It can be very extreme, it can be very subtle. So that's another interesting example. Right. Obviously misinformation, which is another way of saying evaluating the wrong. The model gone wrong is an important component of that. The other thing we do that is that leads to this overall question of like, how do we teach students to think more critically when they're using AI systems? It's just, is to just teach them what is happening in the model itself. So our focus is on the digital sciences. We're very much a, uh, technically oriented education organization. The surface area has grown beyond computer science with everything we've been talking about today, AI and data science. But that technical focus a little bit, going back to our, uh, earlier conversations, I think is an unlock for students in terms of questioning the systems. If you don't understand the system, how do you question it? If you understand how it works, if you understand this domain, the subject matter, you have a better shot of questioning it in a way that is effective, healthy and keeps you on the right side of the lines that we're trying to keep students on, which is high quality information, good decisions, ethical thinking, those sorts of things. So that's what you'll see us do a lot of, is get into the how it works. So when they use it, they know how to direct it, how to steer it, how to create with it makes sense.

Speaker A: And uh, on the other piece of the equation, where we started with teachers, there is a pretty striking point poll recently released by the Walton Family foundation in Gallup that only 18% of K through 12 teachers have received formal guidance from their administrators on how to use AI in the classroom. 34% are getting no guidance at all. So for a teacher listening right now, who feels behind the curve or fearful or unsure of where to start, what do you say to them? What are some of the use cases you've seen of how an educator is using the. The resources you've offered in an interesting, fun, or creative way?

Speaker B: Yeah, the first thing I'd say to the educator who's uneasy, fearful is it's okay. A lot of folks are feeling that way and own that feeling. It's okay to feel that way. It's a very. I mean, folks that are in roles like myself and others, even in the technology industry, are feeling some of those same feelings. Um, things are moving extremely quickly. There's a lot of these are challenging, hard questions that we're faced with. So that's the first thing I would say. I know, I think that educators carry that extra burden of, you know, responsibility for the future of the kids that are in their classrooms. And that's something that I can only. I have some empathy for, but I've never had to stay in front of the classroom year after year and be in that situation. So hopefully the tools and products that we have can support those educators. The thing I would say is you yourself, as an educator, need to learn about this technology. Find your favorite professional learning curriculum or course. We offer professional learning specifically focused on teaching teachers about how this technology works. Now, to think about it in the context of the classroom, there are a number of other organizations, from the A.I. uh, research labs to the big tech companies, they all have versions of learning about AI. Use the one that feels the right for you as a teacher. Right. And some of that depends on are you teaching a 12th grader technical education? Are you teaching, you know, third graders general education that may guide how you think about the type of resource you use. We pair our AI curriculum with very specific professional learning for that curriculum as well. So we have standalone curriculum, learn about AI generally, and then learn about how our curriculum to students is taught. And that pairs really well for, for teachers as well. So that's another thought for your teaching audience, the fun parts. And, you know, I'm probably the last person to be telling the teacher how to have fun with their students. Whenever I go into classroom, I spend time in classrooms. I'm always amazed by just how wonderful our teachers are across the globe. It takes a very special profile and special person to do that work. But what I do see in classrooms When I go in there is teachers who really engage in the work with their students. And I'll say my own experience with a math teacher who took a number of students after school and, you know, spent time with us on a thing that was not math related is an example of that. Like, getting into the work with us. You know, it's sort of like my, uh, analogy for this is I, I visited a classroom that had third grade students, and the teacher got on the floor with the students, like, chin, elbows on the floor, you know, hands on his chin, looking at a laptop with two other kids. And I was like, that's a teacher that understands how to really get into the work. And if there's any moment to do that, it's now because they're learning alongside their students. Right. And that, that ability to do that, I think is powerful. It's powerful for the students as they see that, and it's powerful for the teachers to say, it's okay, I can be a beginner. Again, I say that to my team all the time. We were experts in computer science for a decade. We are not experts in AI science. We are beginners along with a lot of other practitioners. So we have to be curious. We have to get in the work, we have to get into the field. I think when teachers do that, that's when I think they have fun, the students have fun, and hopefully everyone learns.

Speaker A: Yeah, yeah. And the approach of curiosity. I think you've said a few really important things along the way, and just to name two, that everyone should know we're all in the same boat. You know, even the developers of this technology can't answer all the ways in which it works, which is in some ways terrifying. But in some ways it should get everyone, uh, in one agency. You know, we have to bring our different skills and experience to understand what it is we're working with and how it can serve us all. So we are all more in the same boat than we know. And we have to go through this with curiosity. And we really see that as an important piece of the end goal. And I know, Karim, part of the reason we started our conversation in the first place was our shared passion for AI literacy and hope that by offering more AI literacy, more people can feel invited into this AI tent and find more agency. I know when we're talking about it@ equal AI, we don't think that everyone needs to be a computer scientist. It's great if you are. The more you understand, the better. But there are so many different ways we need to approach AI. So all the different disciplines are important to bring to the lens of AI into the discussion. And, uh, I would love for our listeners to understand how do you think about AI literacy? And if we achieved our goal and we had an AI literate global population, what would that look like?

Speaker B: Yeah, I think this is probably one of the most critical questions of this era for us to answer. I have a glib answer, which is, I, I think it's too early to put a label on things because when you say a word like AI, uh, literacy, as an example, depending upon the listener, it can be filled with a definition that looks very different for each listener, each person, given their context. But at the same time, you have to define things to get clear. Right? You have to put. You have to actually say, okay, here's the label, but here are the things that that means. So, uh, I'll start with my highest level definition and get into some specifics that I think are, uh, not just code AIs, but I think there's a community of organizations that center around this definition. So I'll start with my macro view, which is this shift from computer science as a vocational, uh, domain to a foundational science, I think is one that we don't talk about enough. And so I'll use an analogy here. For a thousand years, probably longer than that, probably since the beginning of organized education, we've taught students about the physical world. So we teach them something about biology, something about chemistry, something about physics. They read some books in the US they might have dissected something for, uh, me in middle school, it was a frog. Maybe in physics class or in general science class, they went outside and threw rocks into the air, saw gravity working, and started to really experience the world in a technical way. And that's not because you're a physicist and I'm a doctor. It's because it was important to have a foundational knowledge about how the world around us works. So my perspective and our perspective is we're in a world now where our students are living in a digital world, yet they're not dissecting an AI model. They're not looking closely, inspecting computer programs. We think that's an issue. You know, if they're spending hundreds of hours in all these subjects and only a small number of hours on the digital sciences, how do we prepare them for a digitally oriented world? So that's a macro view that I take about the shift in literacy and what that looks like, getting into the specifics of what literacy looks like. I think there's some Foundational technical knowledge. How do AI systems work? How do computing systems work? How do they make decisions? Getting into even maybe the math underscoring this cross disciplinary right probabilities are a key part of this. So um, getting into the details of what's happening behind the prompt I think is a critical component of literacy, that technical foundational knowledge, a durable skill that I'll highlight going to the next piece. Computational thinking skills, fancy wave saying, learning how to solve problems in the world. But there is a very specific construct called computational thinking that when you apply that to even non computational problems, actually enables more interesting solutions. Teaching coding is not to teach coding, it's to teach a thinking process. And I think in a world with AI, we'll not only teach computational thinking skills, I think we'll go up a step and start to teach a bit more about systems thinking as we work with AI systems that can actually do more. And so things like goal definition, providing constraints, providing requirements, providing evaluation criteria, these are examples of the primitives I'll call systems level thinking. These are durable skills. Again, they're, they're skills that can be applied in almost any context, right? Planning an event, to planning how to grow a crop, to planning, uh, you know, a research run on a new drug, to planning a software program. They can all benefit from what I'll call that durable skill. And then our students like learning about the attitudes is probably the best word to describe it, but in many ways it's sort of attitudes and behaviors that come out of the work that they're doing. We talked about agency, right? That's an important component, right? Giving students that agency. But curiosity is another word that's come up in our conversation. How do we instill that sense of curiosity? Collaboration. These tools, these systems technology requires collaborative engagements, human to human and human to computer, right? So these are some of the attitudes or mindsets that I think are important when we talk about AI. Ah, literacy. So back to your point. The goal is not to create a generation of computer scientists, but I would say it is to create a generation of digitally fluent citizens, just as we've created citizens over the last thousand years that were very physically fluent.

Speaker A: M and uh, it seems from our experience, AI literacy has some important routings in curriculum and tools like what you're offering. And there's also some simple tactics like telling people your favorite prompt and use cases for AI, because you're going to hear some that are going to push people to give it a try and to really understand how it can benefit them as well as how they should not be using it. So in honor of our appreciation of AI literacy by sharing what works, can you tell us, Karim, what is one of your favorite use cases and what is a way you would not use AI?

Speaker B: Oh, that's. I love that. That's a great question. So I like to use AI to kind of sharpen my own thoughts. So at, uh, Code AI, we have a set of principles that we value as behaviors we all try and inspire to, uh, exhibit in our work day, day to day. And one of them is called Sharpen Together. And that's when two people spend time, like going through a problem or a thorny issue, and hopefully through that brainstorming process together, they come up with a solution. And I think that's something that you can do with an AI system as well. It's something that is doable. So even before you take it to the next step, going through a bit of a sharpening process, I tend to use AI systems with a very opinionated perspective of asking the AI system to push back on my idea. So I actually put it in the mode of challenge me, not just in a sycophantic way. Agree with me. And I think what we'll see from some of these AI systems will continue down this path. I think the. The newest 5.0 model from anthropic is an example of that. So I do think these models will continue that way. But I. That's one of my favorite use cases is, tell me where I'm wrong. Tell me what I didn't do right. Tell me about the gaps. What am I missing? So that's the way I like to use it. I caution, I pause on using it to write. Everything I write. I think it's very. It's so easy to just get pulled in by its capabilities around writing. But I don't want to lose my own voice and my own capability of writing and my own skill at writing. So I do use it to do some smoothing of language. But that's probably the place where, for me personally, I've said I'm going to draw a line here and focus more on smoothing than direct writing, because that's especially when I'm writing publicly, I think it's very important to have that genuine human voice and my own voice, the way I speak, that's uniquely me, come through in what I put out into the world. So that's probably a line, at least so far, that I've drawn for myself.

Speaker A: Yeah. Well, it's interesting you say that because that is the routine answer among Our guests. So I'm hoping there are kids out there listening who can understand that it's not a best practice for leaders in this field. They need to learn and develop and make. Maintain their voice. Uh, it's powerful, but we can't let it replace us. We have agency and not letting it replace us in our voice.

Speaker B: Absolutely.

Speaker A: Well. And Karim, I had so many questions to ask you. I apologize. I have buried the lead. We are very excited to celebrate with you, uh, a development coming out this week of our publication of the podcast. I understand you've been doing some important work with the European Commission and the oecd. Can you please tell us more, More about that?

Speaker B: Yeah. So we're thrilled to be partners with the European Commission and the oecd. They have joined forces in a joint initiative to launch an AI literacy framework. It is one of the. I think it is the first of its kind to essentially provide a common, uh, understanding and structure around what primary and secondary education should look like in the domain of AI. Ah, literacy. What we've just discussed, Code AI is a partner of these two organizations. We worked very closely with them to develop this framework, along with actually 100 organizations who provided and from across the globe provided insight input, experts in their fields that helped shape this framework and this guidance. So we're, we're thrilled to be announcing that alongside these two organizations who are really, really leading the charge. And it's going to be a really fantastic resource. It's launching with the idea that we, the group will continue to iterate this forward as this space changes on a very regular basis. But I think the partnership intention is to really affect policy outcomes, learning design and curriculum creation. Educators and teachers thinking about how they form lessons at the very local level in each of the classrooms that they're in. Many teachers are doing their work directly in those classrooms and creating their own curriculum. So at whatever level education is occurring for students, our hopes, our dreams, our desires, is that this framework can be a touchstone for those different ways in which literacy education lands ultimately in the classroom.

Speaker A: Very exciting. Well, congratulations and thank you for that. And, um, I know we have to let you get back to all the many things that you've been telling us about and more that you have underway. So. So a last question we like to ask all of our guests because we're all really in this work because we're excited about AI. I think there are fears and risks and concerns, but I say I'm, um, AI net positive. I think that ultimately, if we get this right, if we're intentional. It is really exciting to think about the opportunities we'll be creating at scale. So let me ask you, Karim, what is something you are excited about right now?

Speaker B: Oh, wow. With respect to AI, I think, you know, the conversations are challenging right now. There's a lot of incoherence. Right now is the best way I'd put it. The space is moving quickly. There's some other factors, you know, factors about the screen time discussions that are occurring in the classrooms that intersect with this work, you know, the broader macro factors about. Is Ann moving too fast impact in communities that, that in current. It's scary. It's hard, it's challenging. I'm, I'm excited to be part of the conversation, help us solve some of these challenges because I think if we actually. It's hard work. It's. These aren't easy days and easy conversations, even this conversation. We've talked a lot about some challenging things that we see in classrooms and, you know, in the pathway to classrooms. But I think it's exciting to be a part of it, to have a voice in the conversation at how lead and to, uh, find collaborators who want the same outcome we do, which is this next generation of students to feel agency, to be empowered, and to feel confident about navigating the world that they are, are living in and will ultimately be, you know, not only, only citizens of, but leaders of, be part of the workforce of and and shape the next generation of, you know, what the world looks like. So I think that excites me. It's a vague, hairy, audacious, scary place to be, but it's also exciting. That's what keeps me motivated.

Speaker A: Well, thank you for sharing that and thank you for all you're doing. Kareem Megji, thank you for joining us today.

Speaker B: Thank you, Miriam, for having me today.

Speaker A: I'm Miriam Vogel with Equal AI and this is in AI We Trust. Thank you for joining us for today's episode. Subscribe or download our podcast on Spotify, Apple Music, Google Play, or wherever you get your podcasts, and we welcome your feedback. If you like the podcast, please rate us, give us a review. And to learn more about Equal AI, visit us at www.equalai. org.