…Children need to be the architects of the future. We are underestimating them. We don’t just need children to be able to use AI tools; we need them to understand them and build the future of these things. That will lead us to a place where children not only feel more empowered, but are able to build a better future because they are the ones in control…
I had the pleasure of talking with Andrew Sliwinski of LEGO Education. Andrew grew up in a household shaped by creativity and making. His parents were, as he puts it, “maker-type people,” and their home was filled with tools and the quiet permission to build, take apart, and experiment. As a child, Andrew was known for dismantling nearly everything he could get his hands on, from toys to computers, often landing himself in trouble. Still, that instinct to understand how things work by using his hands never left him.
That early exposure to creative problem-solving followed him into adulthood and into his formal education. Andrew studied design in Detroit, a city whose public schools would later play a defining role in his career. To support himself while studying, he worked as a tutor. During that time, he was approached by someone from Detroit Public Schools with an offer to help students prepare for standardized math tests. What he encountered in those classrooms left a lasting impression. He saw students who were deeply disengaged, many of whom openly declared their dislike for math or questioned its relevance. “I saw kids in whom I recognized a lot of myself,” he recalls, particularly those who felt alienated by traditional approaches to learning.
For Andrew, that experience marked a turning point. He was young enough, he says, to feel a sudden clarity about what he wanted to do. He became committed to working at the intersection of child agency, creativity, and hands-on learning, a focus that has guided his professional life ever since. That throughline later carried him to the MIT Media Lab, where he co-directed Scratch, a programming language designed to help children learn coding concepts through play and creative expression. It also led to his role on the Board of Trustees for the Raspberry Pi Foundation, an organization centered on learning through making and accessible computing.
Eventually, that same philosophy brought him to his work at Lego Education (part of The Lego Group), where he has helped shape educational products that extend far beyond Lego’s identity as a toy company. Andrew often points to field testing in classrooms as the most meaningful part of that work. One story, in particular, has stayed with him. During a school visit, he and his team arrived to test a product, only to learn that the school had quietly separated students. Those labeled as remedial had been placed in a different classroom, leaving only higher-performing students to participate.
As the lesson unfolded, teachers began to realize something unexpected. The hands-on activities were not just effective for advanced students, but deeply engaging for all children. Midway through the session, teachers ran down the hallway to bring the other students back. What they saw, Andrew says, was that the learning experience could bridge gaps rather than reinforce them. That pattern has repeated itself across classrooms, including among students with dyslexia, autism, ADHD, and physical disabilities. “If you engage kids in a hands-on way around things that they care about, they will show up,” he says.
Much of Andrew’s work has focused on using play to introduce complex ideas in science, computer science, and artificial intelligence. One science lesson he often describes involves a playful scenario called “Lemonade Shake,” where students build a lemonade stand and simulate an earthquake. As Lego minifigures watch their lemonade spill, students work together to engineer structures that can withstand the shaking. Beneath the silliness, Andrew says, students are learning engineering principles, earth science concepts, collaboration, and empathy, all within a single class period.
More recently, his work has expanded into computer science and AI education. In one lesson designed for elementary school students, children train a simple machine learning model to recognize dance moves inspired by popular culture. Working in groups, they then build and program small robots to perform those dances. Through laughter and movement, students encounter ideas like probability, data quality, and model training. Andrew notes that this approach often reaches students who might otherwise see computer science as uninteresting or exclusive.
A central concern in this work has been the role of teachers. Andrew emphasizes that educational tools must be designed not only for students, but for educators who may feel uncertain about teaching emerging subjects like AI. He cites research suggesting that about half of computer science teachers remain uncomfortable teaching AI even after training. For that reason, his team tests products by handing them directly to teachers and observing from the sidelines. If teachers cannot confidently lead a lesson, Andrew argues, even the best-designed technology will fail to scale.
Underlying all of this is a belief in open-ended learning. Rather than aiming for a single correct answer, Andrew’s approach encourages what he calls “solution diversity,” allowing students to arrive at many valid outcomes through collaboration and discussion. He sees this as essential to developing critical thinking, problem-solving skills, and social awareness.
Andrew is also a strong advocate for screen-free learning, particularly for younger children. While much computer science education relies heavily on individual screen-based tasks, he believes many foundational ideas can be explored more effectively through physical materials and group work. Screens, in his view, should be an extension of learning, not its starting point.
Looking ahead, Andrew hopes to shift how adults think about children’s relationship to technology. While much public attention is focused on what AI systems can do, he believes society has underestimated what children are capable of. “Children need to be the architects of the future,” he says. For him, the goal is not simply to teach children to use powerful tools, but to help them understand how those tools work, so they can shape what comes next.
Yitzi: Andrew, it’s an honor to meet you. Before we dive deep and talk about Lego and education, our readers would love to learn a bit about your origin story. Can you share with us the story of your childhood, how you grew up, and the seeds for all the amazing things that have come since then?
Andrew: It’s a very winding journey. When I tell the story to people, it all checks out and makes sense. It seems like that all naturally led together. However, in the moment, it felt very winding. I have parents who are incredibly creative, maker-type people. Growing up, I was really lucky to be surrounded by tools, but also that spirit of building and creating things. I was one of those kids that took everything apart. I took apart my toys, the computer — I took apart everything, including getting into a lot of trouble.
That spirit of building, discovering, and using my hands has followed me my entire life. I was lucky to go to school in Detroit to study design. While I was there, I worked as a tutor to help pay the bills. One day, someone from Detroit Public Schools came in and offered me the opportunity to go to the schools to help students with their standardized math tests.
I saw kids in whom I recognized a lot of myself — kids who were completely disengaged with math. It’s kids saying things like, “I hate math,” “Math hates me,” “This is the worst subject ever,” or “How am I going to use this?” All of that really resonated with me. There was this moment where I was lucky to be so young and have that sense of, “Okay, this is what I want to do.”
From there, I’ve had a history of always trying to work at that intersection of child agency, creativity, and hands-on engaged learning through everything I’ve done since. This includes my time at the MIT Media Lab co-directing Scratch, which is a programming language for children. I sit on the Board of Trustees for the Raspberry Pi Foundation, which is connected to this idea of learning through making. And then last, but certainly not least, my work at the Lego Group.
Yitzi: Amazing. You probably have some amazing stories working at Lego. Can you share with our readers one or two stories that most stand out in your mind? Or to reframe it, if you were to write a memoir, what would be the most important story you need to have in it?
Andrew: Some of my favorite stories of designing and building products at the Lego Group come from field testing. There is a great story where we were testing one of the products. When we got to the school, the school had decided, unbeknownst to us, to separate the classrooms. They took all the remedial students — the students that struggle in school more than others — and pulled them out of the class and put them in a classroom down the hallway. They just gave us the non-remedial students.
We started the lesson, and there was this moment of realization from the teachers where they realized, “Oh, this is not just great for our advanced students. It’s great for every kid.” They literally ran down the hallway to grab all the kids from the remedial class to pull them back into the classroom. They saw how what we were doing was not just for the kids who were excelling in school, but actually helped bridge the gap.
That is something you see with almost everything we do. If you engage kids in a hands-on way around things that they care about, they will show up. We see this with dyslexic students, ASD students, ADHD students, and students with disabilities. We are able to bridge across student populations in a unique way that I would love to see spread across more of education. That moment of a teacher realizing that we are able to help them bridge that gap will always really stick in my mind.
Yitzi: That’s amazing. Let’s move to our main subject. When people think of Lego, obviously it’s an amazing, extremely successful toy. It’s not intuitive that it could be used as a teaching or learning tool. Aside from maybe architecture, can you give some more basic examples of what different subjects could be taught with Lego?
Andrew: Last year, we launched a science product. I have too many favorite science lessons from that product because it covers kindergarten all the way through eighth grade. There are over a hundred lessons. One of my favorites is called “Lemonade Shake.” It involves the idea where kids work together in a group of four to build a lemonade stand and an earthquake simulator. The minifigure has their lemonade stand, and when the earthquake happens, all of their lemonade flies everywhere. The kids start laughing; it’s really silly.
Their task is to use Lego bricks to construct an earthquake-proof structure that protects all the lemonade and keeps it from splashing around. In that simple lesson, which connects to younger kids’ innate sense of humor, they are learning all about engineering and earth sciences. They are connecting empathetically to these little minifigures that they want to protect. They are working together in a group of four, so they are learning all about collaboration. We are able to hit so many different dimensions of learning in this single 45-minute lesson that, on its surface, is very silly. Every time we go to a school and see teachers test with that lesson, there’s an “aha” moment: “I see the learning. I feel the learning. I care about the learning.”
With the product we are launching today around computer science and AI, we are able to build on some of those same ideas where children develop an understanding of very deep AI concepts. This is not just about using off-the-shelf AI tools, but going deeper and getting to AI fundamentals.
One of my favorite lessons for third through fifth graders involves something kids are really interested in these days: dance, connecting to TikTok and Fortnite. We have kids work together in a group of four to train their own machine learning classifier to identify different dance moves. Each kid contributes their own dance moves. Then, they build a little robot using motors that can show off different dance moves. They program this to replicate the dance moves that they just built their machine learning classifier to identify.
Here we are getting to complex ideas about computer science and AI. We are getting into probability, training your own machine learning model, and data quality — garbage in, garbage out. We get to all these big concepts through developing a dance party in the classroom. You see all this joy and engagement, particularly from kids who think computer science and AI are for “dorks” or “nerds.” They see the dance, the silliness, and the engagement with classmates, and we connect them to very fundamental artificial intelligence concepts.
Yitzi: So the idea is that they train the Lego robot to dance?
Andrew: Yes. And that’s just one of many lessons. This represents 45 minutes of over 160 hours of learning materials we’ve built. It is a great, fun example. I particularly like to see that one when we go out for testing because every classroom, every culture, and every school does it a little differently.
Yitzi: That is so smart. Is the vision that this should be deployed in classrooms across the world? Or is it that parents will use it with their kids?
Andrew: We design first and foremost for schools, imagining your standard classroom. We also design for teachers who are often really uncomfortable teaching computer science and particularly AI concepts. We know that roughly 50% of all computer science specialist teachers feel uncomfortable teaching AI even after they’ve had training. That’s a pretty striking statistic.
When we design these products, we need to design them not just to deliver the learning outcomes for the students, but for the teachers so that they feel confident bringing these ideas into the classroom. When we test these products around the world, we aren’t the ones teaching. We hand the product to a teacher and say, “Here are the basic materials. We’ll observe, but you are the teacher.” We hold ourselves to that bar because if you build the best product in the world, but a teacher doesn’t feel comfortable teaching with it, it’s not going to work. Our priority is to make sure these things work at scale in schools and classrooms around the world.
Yitzi: Unbelievable. Could you teach math with these?
Andrew: You can. There are a lot of mathematical concepts in computer science and AI, but we really specialize and focus on specific curricula and academic standards. For our science product in the United States, we map everything to the NGSS (Next Generation Science Standards). For the CS and AI product, those 160 hours of learning map to the CSTA (Computer Science Teachers Association) standards.
The best way to build great learning experiences is to get hyper-focused on what those learning outcomes are. In CS and AI, there are many mathematical concepts, with statistics and probability being the most obvious. We touch on those concepts, but we don’t have any products that are just focused on mathematics today.
Yitzi: That’s great. How can this curriculum be used to stimulate critical thinking?
Andrew: When children engage with any of these lessons, there’s always a part where they are introduced to a concept. That could be probability, or something as simple as a loop or a variable. Secondly, they are generally introduced to a problem or challenge. They have to work together in a group of four to rise to that challenge. The way we build everything ensures there is not just one right answer — which is very different from almost everything else in school. In everything we do, there are an infinite number of right answers; we just help the students arrive at them. We call that “solution diversity.”
There is so much critical thinking that happens in that group of four students. They have to navigate relationships and mediate discussions or disputes to arrive at a solution. That is not just helping them develop critical thinking and problem-solving skills in the abstract; it is very active.
There are many different “secret sauces” here. One is having open-ended problems. One is supporting solution diversity. But the third is that collaborative element. The students aren’t just developing these critical thinking skills in isolation; they are developing them in a collaborative context.
Yitzi: That’s great. Maybe this is slightly afield, but I think it’s related. Can you articulate the importance of screen-free play for children?
Andrew: Absolutely. To focus on computer science and AI, if you go into most classrooms where these are being taught, the predominant way students learn is by sitting in front of a screen by themselves, often with headphones on. They are often working on an incredibly constrained set of tasks — I call them “maze solvers” — trying to get a character from point A to point B. They do that again and again. They try to deal with student motivation by making the characters cute, but it is still a constrained set of problems that hits on a small set of standards and limited child interests.
I think it is critical, particularly for younger students, to realize that so many concepts in computer science and AI don’t need a screen to be taught or explored. Going back to my earlier point, it’s so much stronger if they do it in a group. A lot of what we’ve done in the product we’re announcing today involves screen-free exercises where children can build and explore computational thinking concepts without a screen. It is critical to provide that continuum. Often, screen-free activities are how students get started. Then, you provide the possibility of connecting to a screen to take it to the next level.
That is such a different vision for computer science and AI education than the predominant models we see today.
Yitzi: Amazing. This is our signature question. You’re building an EdTech solution. Based on your experience, can you share the five principles needed to create a successful education technology?
Andrew: That is a great question. Only five? [Laughs]
- First, it really needs to be child-centered.
- Second, it must be teacher-centered. I think number one and number two have to come together. Often, educational technology focuses on one or the other. You can build something students love, but if it doesn’t work for teachers, it will never make it into the classroom. Conversely, if teachers love it but kids hate it, that won’t scale either. That balance is essential.
- Third, never let the tail wag the dog. By that, I mean never let the tech lead the learning. The learning should always lead the tech. Particularly with AI, we see examples every day of the tech leading the learning, where the pedagogy is secondary. That is a missed opportunity. You must ask: What do you want children to learn? And then, what does it take to get them there?
- Fourth is collaboration. Every time we test in the field, teachers say, “The tech is cool,” or “The models are cute.” But what always jumps out is when they say, “The kids never get to work together like this.” That is incredibly powerful. If you were to list the most important skills for children in the year 2030, collaboration would be near the top. We have such a focus right now on personalized learning, which isn’t bad, but it needs to be balanced with seeing children as part of a society. If we want to set children up to be the architects of the future, they need to know how to work together.
- Fifth is leaving space for child agency and creativity. Children want to express themselves. They have strong ideas about AI, science, and the world. We often forget that. Educational technology often doesn’t leave space for child expression. Along with collaboration, this is critical. Children want to feel a sense of agency over their own learning. Think about your favorite classes or teachers; they made you feel like you were a part of your learning.
Yitzi: Amazing. This is our final aspirational question. Andrew, if you could spread an idea or inspire a movement that would bring the most amount of good to the most amount of people, what would that be?
Andrew: Right now, we as adults are probably — understandably — obsessed with what computers are capable of. Every day seems to bring some new milestone in AI that fills us with incredible positivity, deep existential dread, and everything in between. But in that obsession with what computers are capable of, I think we’ve lost track of what children are capable of. We’ve become obsessed with the idea that AI is an unstoppable tidal wave coming to erase human relevance, and our job is to help children learn how to swim.
That misses something incredibly important. Children need to be the architects of the future. We are underestimating them. We don’t just need children to be able to use AI tools; we need them to understand them and build the future of these things. I would love to see a movement where we stop focusing on chasing the latest frontier model and catch our breath to help children understand how these technologies truly work at their core. That will lead us to a place where children not only feel more empowered, but are able to build a better future because they are the ones in control.
Yitzi: Unbelievable. How can our readers learn more? If any teachers are reading this, how could they bring this to their classroom?
Andrew: Go to education.lego.com. We have all sorts of different things going on there. For parents and teachers, I would love to see our products in more schools, but I also think we need some of these ideas to make it into schools as well. The values I shared — child-centricity, making things work for teachers, not letting the tail wag the dog, and focusing on collaboration and creativity — are vital. Parents and teachers have a powerful voice in their children’s schools. Advocating for these things on behalf of children can be more powerful than people realize. I would start there.
Yitzi: Andrew, thank you so much for this extremely thoughtful and enlightening discussion. I wish you continued success and good health.
Andrew: It’s really a pleasure.
Yitzi: Have an amazing day my friend.
Andrew: Thanks, you too. Take care. Bye.
Andrew Sliwinski of LEGO Education on Why Children Should Be the Architects of the AI Future was originally published in Authority Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.