I was thrilled recently when my friend Josh Brake mentioned me in his Substack post about “The Forward Deployed Educator.” He referenced the Unit Circle Demo I had created and wrote about how educators can now use AI tools to build custom learning experiences for their students. What made me smile was that Josh included links to technical resources and coding workflows that were honestly way beyond my understanding—Railway deployments, Claude Code workflows, advanced engineering examples. I’m not a coder. Never have been.
But that’s precisely the point, and it underscores two critical insights about working with AI:
First, this is a genuine superpower that AI has given all of us. You don’t need to be a software engineer to bring your pedagogical ideas to life. You need domain expertise, curiosity, and the willingness to iterate through conversation with AI.
Second, and crucially, this superpower is deeply dependent on domain knowledge. More than once, I’ve accepted what AI generated at face value, only to realize later that it had led me astray, sometimes subtly, sometimes significantly. I had to go back to first principles, to my understanding of mathematics, physics, or pedagogy, to verify accuracy and catch errors. AI is powerful, but it’s not a substitute for expertise. It’s an amplifier that works best when guided by someone who knows what “right” looks like.
This creates what I’ve written about elsewhere as a kind of Matthew Effect: AI makes the rich richer. Moreover, as I have written elsewhere, AI also adds to experts’ cognitive load. Those with deep domain knowledge can leverage these tools to become even more productive and creative, at a cost. But for novices (students, early learners, those still building foundational understanding) AI poses real dangers. Without the conceptual scaffolding to question its claims, and faced with AI’s tendency toward sycophancy, learners can be led confidently down wrong paths, through a process of conversational drift. The very fluency that makes AI compelling also makes its errors more insidious. This is why these tools, powerful as they are, must be used alongside, not instead of, the hard work of building genuine expertise.
With that in mind, here are the interactive projects I’ve been building over the past few months, building through conversation with AI rather than traditional programming. Typically these projects start with curiosity, a “what if?” moment, and unfolds through play and iteration. No grand plan, just following where the idea leads.
These experiments span things I care about: making abstract concepts visible, reconnecting with my heritage, challenging misconceptions, and sometimes just building things because they seem fun. Here’s what I’ve made so far.
Making the Invisible Visible
Science often asks us to imagine what we cannot see—waves interfering, gases expanding, bacteria evolving. These simulations make those invisible processes tangible, turning abstract phenomena into interactive experiences you can manipulate and observe.
Interactive Variation on The Way of Code Inspired by Rick Rubin’s creative philosophy, I built an interactive version with sliders you can manipulate, treating coding itself as a creative act.
Wave & Interference Simulations – Exploring how waves interact and patterns emerge:
Flocking Birds – Watch complex group behaviors emerge from simple rules. A beautiful example of how local interactions create global patterns.
Bacteria-Antibiotics Simulator – Models bacterial populations under antibiotic pressure, showing how resistance develops over time.
Gas Laws Simulations – Interactive explorations of how pressure, temperature, and volume interact:
Physics Misconceptions Simulations – These deliberately challenge our intuitions about how objects move:
- Heavy vs Light Ball Drop
- Push vs Drop
- Running Ball Drop
- Blog post on using incorrect simulations as teaching tools
Grass, Cattle, Carbon: An Adaptive Multi-Paddock Grazing Simulator — This interactive simulation grew out of a conversation with Peter Byck, Professor of Practice in Journalism and Mass Communication at Arizona State University and director of the Carbon Nation documentary. During a workshop, Peter suggested building a hands-on tool to help people grasp how Adaptive Multi-Paddock (AMP) grazing works, something I knew nothing about. But here it is…. Adjust herd size, paddock count, grazing duration (down to sub-day moves), rainfall, and soil type, then watch how these choices ripple through forage biomass, soil health, and carbon sequestration over seasons and years. Every key term has a hover tooltip, making it as much a learning tool as a simulation.
The Visual Fourier Transformation: This simulation demonstrates how the Fourier transformation can recreate any drawing using chains of spinning circles, each rotating at its own speed and size. It’s a beautiful blend of math, art, and vibe coding that makes a foundational concept in signal processing visually intuitive.
Mathematics in Motion
Numbers and equations can feel static on a page. These projects bring mathematical concepts to life through movement, color, and interaction—making trigonometry tangible and exposing the difference between correlation and causation.
Unit Circle Simulators – Making trigonometry visual and tangible:
- First version with blog post documenting the process
- Improved version – smoother, more intuitive
Correlation is Causation – A satirical demonstration using real data to show how easily we confuse correlation with causation. The statistics are real; the causal claims are absurd. Perfect for sparking conversations about critical thinking. (Related reflection on AI reasoning.)
Culture & Expression
Code can be a form of cultural expression—a way to honor heritage, explore rhythm, and play with language and form. These projects are deeply personal, connecting technology with identity, music, and art.
Learning Odia – The most personal project on this list. An interactive tool helping me reconnect with my heritage by learning to read Odia, the language of my childhood.
Tabla Teen Taal Simulator – An interactive way to experience one of the foundational rhythms in Indian classical music. A personal bridge between my cultural heritage and technology.
Artistic Typeplay – Playful experiments with typography and design through code.
Tools for Play & Learning
Not every project needs to solve a problem. Sometimes it’s about adding delight, surprise, or a different lens through which to see the world. These tools invite exploration and play.
MCU Timeline – An interactive map of the Marvel Cinematic Universe. Because sometimes you just want to visualize how all those stories connect.
691 Randomizer – Built for my DCI691 course to inject spontaneity and surprise into classroom activities.
Each project started with curiosity and unfolded through conversation with AI tools. Some worked immediately, others took multiple iterations. But make no mistake: every single one required my domain knowledge to verify, correct, and refine what the AI produced. I had to check the physics equations, verify the mathematical relationships, ensure the pedagogical framing was sound, and catch the subtle errors that would have undermined the learning experience. The joy is in the exploration—seeing ideas take shape, testing what’s possible, and learning by making. But the quality comes from bringing expertise to bear on every iteration, questioning what the AI generates, and knowing enough to recognize when something is wrong.
More experiments to come—always guided by the knowledge that makes them worth creating.
The Deeper Point
Back in 2023, after creating my first interactive simulation with ChatGPT, I wrote something that feels even more important now:
“First, lets take a moment to just marvel at what I could create in just a few hours of working with ChatGPT3. I have absolutely no knowledge of Python, or even how to use Terminal on my Mac. But here it is. This is something a middle schooler can create. Just let that sink in.”
A recent conversation with Danielle McNamara helped clarify something I’ve known but perhaps not articulated clearly enough: The point of these experiments isn’t primarily about what I, or any educator, can create FOR students. The deeper pedagogical opportunity is what happens when students create these things THEMSELVES.
When students build simulations of wave interference, or bacterial evolution, or gas laws, they’re not just making a tool, they’re externalizing their understanding. They are creating what Perkins would call a “performance of understanding.” And when they work in groups in creating these artifacts they are instantiating what Shulman said:
“Learning is least useful when it is private and hidden; it is most powerful when it becomes public and communal. Learning flourishes when we take what we think we know and offer it as community property among fellow learners so that it can be tested, examined, challenged, and improved before we internalize it.” — Lee Shulman
0 Comments