Uri Wilensky, Northwestern University’s Lorraine Morton Professor of Learning Science and Computer Science, recently won the 2025 Yidan Prize for Education Research. The $3.8 million prize, widely recognized as education’s Nobel Prize, was awarded for Wilensky’s work developing NetLogo—the most widely used agent-based modeling environment in education. (If you’re uncertain about just what that means, read on.) The Yidan committee noted that Wilensky’s work looms especially large in the age of AI. Intrigued, I reached out to learn more about what it means for students and educators. Here’s what Wilensky had to say.
—Rick
Rick: Uri, congratulations on winning the Yidan Prize for your work on agent-based modeling! But I suspect I speak for many when I say I’m unsure about just what this means. Can you explain?
Uri: Thank you, Rick. My work focuses on how computer-based representations can help people learn. We know that different ways of representing the same knowledge can have a major impact on thinking and learning. For example, Roman numerals and Hindu-Arabic numerals both represent numbers, but multiplication and division are exceedingly difficult to do with Roman numerals. The adoption of Hindu-Arabic numerals in Europe, beginning in the 11th century, made it possible for even children to learn multiplication and division, a tremendous educational achievement. Hindu-Arabic numerals also led to numerous advances in mathematics, science, and commerce. Similarly, I argue that computer-based representations can lead to totally new ways of understanding complex systems, making them accessible to more people.
Rick: What do you mean by a “complex system?”
Uri: Complex systems are those composed of many interacting elements from which larger patterns emerge. For example, the world economy is a complex system emerging from a great number of financial transactions among people and institutions. Complex systems are increasingly recognized in the natural and social world we live in. This requires a new educational paradigm that prepares people to analyze and make sense of complex systems. Computers are an essential tool for this. They can be used to create models that students can interact with to gain an understanding of how these systems work.
Rick: So, what does work on “agent-based modeling” entail?
Uri: Agent-based modeling (ABM) is a form of computer modeling where elements of a system are represented by computational objects called “agents.” Agents can be anything from trees in a forest to people in a city to atoms in a molecule. Each agent has properties such as location, size, and color. Critically, they also have rules governing their interactions with other agents. For example, when modeling a gas in a container, you can model the molecules as balls, each of which interacts according to the following rule: If it encounters the sides of the container, it follows the “bounce rule,” and when it encounters another molecule, it follows a separate “collide rule.” This kind of modeling can also be applied to social phenomena. Once the rules are set, you start the simulation and observe how the system behaves over time. Because ABMs are composed of simple agents and rules, and their assumptions are visible, students can change those assumptions and see how the behavior of the model changes. This makes ABMs different from most of the computer models we encounter in daily life—such as weather, traffic, or disease models—where we see the outputs but not the underlying assumptions.
Rick: Why is this work so important for education?
Uri: I am part of an educational tradition that emphasizes learners as active creators of knowledge. My lab at Northwestern University—the Center for Connected Learning and Computer-Based Modeling (CCL)—partners with teachers to design learning materials that allow students to explore, modify, and construct ABMs. Then, we study how these materials impact student learning of complex systems. The CCL also works with scientists and policymakers to model real-world systems that can then be used in areas such as public health and environmental policy. This work helps students recognize the power of these models to explain the real world.
Rick: The Yidan citation paid particular attention to your work creating NetLogo. What is that?
Uri: NetLogo is a computer language and software platform I designed for creating and running agent-based models. It was designed with a philosophy of “low threshold, high ceiling,” a phrase emphasized by my dissertation adviser at MIT, Seymour Papert. The goal is to make learning environments easy to get started with and yet capable of powerful computations. In the context of NetLogo, this means it is accessible to young kids to build and explore ABMs, and also useful for scientists conducting advanced research. I believe we have been successful on both fronts: Thousands of classrooms and curricular units at all grade levels use NetLogo, and thousands of published scientific articles have used it as well. This approach is part of why it is the world’s most widely used ABM platform.
Rick: Why is NetLogo important for students and educators?
Uri: NetLogo comes with a vast library of ABMs that can easily be incorporated into lessons, enabling students to engage with models in almost every subject. For example, NetLogo can help earth science students study forest fires. They might start with a simple fire model that has one input parameter: the density of the forest. By running the model at various density levels, students see that beyond a certain threshold, the fire suddenly goes from spreading just a little to a raging fire. Our intuition may be that a little more density would lead to a little more fire spread. But that’s not the case here. This helps students understand a property of many complex systems: They have “tipping points.” Once the students discover this, they may have further questions. What happens if there’s wind or hilly terrain or different kinds of trees that ignite at different temperatures? The students can insert new agents or rules into the model to test how these conditions affect the spread of fire.
Rick: The Yidan Prize judges deemed your work especially timely given the emergence of AI. Why?
Uri: There’s a danger that students will frequently use large language models (LLMs) to get rote answers to questions. With ABMs, on the other hand, students have to actively think about how the phenomenon should be modeled or what the main causal elements are. Our approach engages students in thinking and reasoning about models. This is a very active use of their brains. This is aligned with my belief that, in order to prepare students for a complex world, we should be teaching students to use their own reasoning and creativity to solve complex problems. LLMs, and AI in general, could be of great benefit in education, but they need to be used as a thinking partner for students, not a thinking replacement.
Rick: Obviously, we’re in a moment marked by significant concern about the harmful effects of screens and cellphones. How does that square with your work?
Uri: My viewpoint is that the harmful effects don’t come primarily from using screens but rather from how they are used. Passively consuming social media content is rarely a way to engage with powerful ideas. We need to design learning experiences that help students understand and approach complex, real-world problems through hands-on exploration, construction, and discussion. I believe actively engaging with models is one of the best ways to learn powerful ideas and foster critical thinking.
Rick: If you’ve one insight that you most want to share with educators, what would it be?
Uri: In order to understand the world, we create mental models of what we perceive. When we encounter ideas that don’t fit into our mental models, we find them difficult to understand. By acquiring a broader set of mental models, we can make new concepts easier to understand. An example is the prisoner’s dilemma, a thought experiment in which two people must choose between cooperation and betrayal without knowing what the other will do. While betraying a partner often offers a better individual reward, if both partners choose to betray, they both end up with a worse outcome than if they had cooperated. This concept can help us understand collaborative and competitive dynamics in different settings. Additionally, since we live in an increasingly complex world, the number and complexity of models needed to make sense of it will necessarily increase. It is therefore vital to equip students with the skills to create, use, analyze, and critique agent-based models. These skills are essential to creating a modeling-literate society.
This piece has been edited for length and clarity.
