Rolling for Insight: How Gen AI and Dungeons & Dragons Are Built on Collaboration and Creativity
At first glance, generative artificial intelligence (AI) and Dungeons & Dragons (D&D) seem like they belong to entirely different worlds. One is driven by algorithms and data; the other, by epic quests, magic, and imagination. But here’s the thing — they both hinge on creative problem-solving combined with the power of teamwork.
Over the past year, I’ve been teaching basic AI skills through general frame of the highly entertaining world of D&D, and it turns out the two aren’t as far apart as you’d think. Whether it’s outsmarting a cheese-obsessed warlock or refining an AI model that keeps reinforcing bias, D&D gives us a way to explore the very same skills we need in AI: critical thinking, collaboration, and pushing the boundaries of what’s possible.
So, let’s gather our party, roll initiative, and explore how bias, moral reasoning, rich context, and creativity link these two worlds in surprising ways.
Game-Based Learning: Teaching AI Through Dungeons & Dragons
When it comes to teaching AI through D&D, the approach fits squarely under Game-Based Learning (GBL). GBL uses actual games to immerse learners in an environment where learning is intertwined with play. In these workshops, D&D isn’t a teaching tool slapped onto traditional lessons — it’s at the core of the learning process itself.
Instead of just reading about algorithmic bias in a textbook or doing dry prompt engineering exercises, learners experience it directly by playing through scenarios. For example, when AI generates characters based on biased training data, it mirrors how we might default to familiar, stereotypical roles in a D&D campaign. The gameplay becomes the lesson — whether they’re deciding if their tiefling rogue should trust the orc wizard who hoards cheese wheels or whether they should challenge the AI’s output to break free from its initial outputs.
Game-Based Learning works because it taps into the natural engagement that games provide. In D&D, players are drawn into the story, the quirky characters, and the world they build together. They don’t just learn concepts like fairness or ethics in system design — they live them, solving problems collaboratively and reflecting on the decisions they make. The learning is embedded in the gameplay, making abstract ideas like “bias in algorithms” tangible, memorable, and fun (especially when a cheese-loving dragon enters the scene, or in a recent workshop, an entire world of potato-themed super warriors).
Unlike gamification, where game-like elements are tacked onto a non-game context, GBL makes the game itself the learning environment. By integrating basic AI concepts into D&D’s deep narrative structure, learners don’t just study AI — they experience it.
Character Creation: Probing Bias in AI Outputs
One of the first things you do in any D&D campaign is create your character. It’s a chance to imagine someone or something entirely new — a gnome bard with a hatred of dairy, a dragonborn wizard moonlighting as a cheese connoisseur, or perhaps a halfling with a cheese-detecting pet pig. The possibilities are endless. But what happens when you use an AI tool as a collaborator for character creation? (I even built a custom GPT for these purposes that also sometimes responds in riddles and epic poetry as I am easily entertained.)
Here’s where things get interesting: AI, just like us, has biases. Much like how D&D players tend to fall back on familiar archetypes (the ever-present brooding rogue with a tragic past, probably named Kael Ironborn), AI models rely on the patterns embedded in their training data. When you ask AI to generate a character, does it automatically suggest a white male knight or an elegant elven princess with long flowing golden hair? What if you ask for something different — like an older, pudgy, gender-nonconforming half-orc, half-elf who collects rare cheeses? How long do you need to prompt it before it moves past its default responses? How much do you have to prompt, re-prompt, push, and start over to guide the AI tool where you want to go?
D&D pushes us to stretch our imaginations. AI…not so much — unless we actively “kick the robot” into thinking beyond its defaults. By questioning why the AI keeps giving us the same characters, we begin to recognize the biases built into its design. When we challenge these defaults, we expose the biases shaping the AI’s world — and ultimately, the systems we create.
Building the World: Setting the Stage for Interactions
In both D&D and AI, the real magic lies in complexity. Dungeon Masters (DMs) build entire worlds filled with history, conflict, and unexpected details — like taverns run by goblins with secret cheese-making empires or dragons who hoard cheddar instead of gold. Similarly, when learning to use AI tools, users actively engage with complex environments where their decisions and actions are shaped by the system’s underlying data, algorithms, and structure.
These seemingly small details — like a dragon’s obsession with cheese — can have a major influence on how players interact with the world and make decisions. A cheese-hoarding dragon might tip the party’s focus toward finding rare cheeses instead of treasure, shifting the entire course of their adventure. This parallels how minor biases in AI systems can lead to unexpected or disproportionate outcomes. A slight quirk in the data or an overlooked variable can significantly influence the direction of an AI system’s decisions, just as a cheese-loving dragon might change the fate of an entire campaign. (Generative AI encouraged me to make the cheese examples consistent throughout to engage the reader, so here we go. Please take all overuse of cheese-related-examples complaints to ChatGPT.)
Whether it’s a brie-obsessed villain in D&D or an AI model that inadvertently favors certain outputs due to biased data, small quirks in world-building shape the decisions and outcomes of both games and systems. The more inclusive and thoughtful the worldbuilding, the richer the experience becomes. In both contexts, learners or users are encouraged to collaborate, experiment, and reflect on their choices, leading to deeper and more meaningful interactions.
Rolling for Initiative: Collaboration and Teamwork in AI and D&D
Once your characters are created, the adventure begins. Maybe your party has to liberate a village from a cheese-hoarding dragon or outwit a guild of sentient mice in a negotiation gone wrong. No single character can handle these challenges alone — it takes teamwork, drawing on each member’s unique strengths, whether it’s the rogue’s stealth or the bard’s ability to charm the cheese merchant into giving a significant discount.
Learning to use AI tools, especially in larger projects, works in much the same way. No one person can master every aspect of complex AI-driven projects. It takes a collaborative team — students, creatives, data enthusiasts, ethicists and more — each bringing their strengths to the table. Just like a well-balanced D&D party, different perspectives are essential for exploring AI’s potential in creative projects, problem-solving, and education.
Hollander (2021) notes that D&D fosters moral experimentation through collaborative storytelling. Similarly, learning AI within a team setting encourages reflection on ethical issues like fairness, transparency, and bias. Collaboration is key — not just for using AI tools effectively, but for ensuring thoughtful, ethical outcomes. Much like a D&D party outwitting a dairy-obsessed villain, AI users must work together to navigate challenges, ask pointed questions, and listen for understanding, making sure the final project is rich with diverse insights and creative solutions.
Immersed in the Game: Entering the Flow Zone
One of the best things about D&D (and, let’s be honest, AI) is that moment when you’re so absorbed that you lose track of time. You’re so deep in the game (or project) that hours seem to disappear. This is called flow, and it’s when you’re totally immersed in the task at hand.
In D&D, flow happens when the party is busy strategizing how to sneak past the dairy-obsessed minotaur guarding a vault of rare cheeses, or they’re figuring out how to outsmart a wizard with enchanted cheese wheels. (So much cheese at this point, so much cheese…) In collaborations working with generative AI, flow happens when the team is fully immersed in problem-solving, debating the best approach to optimizing outputs, exploring new processes, or prototyping differing solutions. The task is challenging enough to keep everyone engaged, but not so difficult that it becomes overwhelming.
So, whether you’re in the middle of an AI project or a D&D campaign, flow is where the real breakthroughs happen. It’s where learning becomes second nature, and the magic — cheese-loving creatures and all — starts to come alive.
Victory! The Magic of Collaborative Problem-Solving
Whether you’re slaying a dragon that hoards gouda or solving an AI bias problem, victory is never about one person. It’s always about the team — how everyone worked together to achieve the goal. The bard’s charm, the rogue’s sneak attack, and the wizard’s cheesy fireball all played a role. The same is true in AI. The coder, the designer, the ethicist, the creative, the learning designer, the subject matter expert — they each contribute something critical to the project’s success. We are stronger when we deeply collaborate and learn to view the world through another’s eyes, even if just for a few minutes.
D&D teaches us that collaboration is the key to success. AI projects, while more subtle about it, operate on the same principle. As Hollander notes, D&D encourages empathy and growth by making players confront ethical dilemmas as a group. The true lesson in both D&D and AI is that it’s not just about technical skills or clever strategies — it’s about how well we can listen, adapt, and work together.
So, whether you’re navigating a cheese-filled dungeon or debugging code, the journey is always bigger than the problem itself. It’s about challenging assumptions, pushing boundaries, and creating something unexpected. That’s where the real game begins.
