06 · Types of Interactive Narratives | Generative AI for Game Development
This chapter companion maps six types of interactive narrative, each with an example game, a cost and a role for generative AI. It ends with an AI suitability comparison and a rule to choose structure before deciding what the model may do.
This video is a companion to the sixth chapter of Generative AI for Game Development by Shesh Narayan Gupta. The previous chapter covered how stories are structured, and this one covers how the player meets the story moment to moment. It maps six delivery systems for interactive narrative and shows which ones generative AI supports best. Its starting point is that dialogue choices are only one delivery system.
Branching narratives
At key moments the player chooses and the story splits. Some branches rejoin the main road, while others lead to a different ending. The Walking Dead by Telltale tracks every lie, mercy and abandonment and returns them episodes later. This suits narrative RPGs and moral dilemmas, and the price is that writing multiplies with every fork. AI helps by generating variations, alternate dialogue, consequences and scene rewrites, and by filling gaps between branches so the story stays coherent.
Open-ended narratives
In an open-ended game there are no chapters or missions, only a world, and the story arrives in fragments such as a rumor, a ruin or a stranger's warning. Skyrim is the example, with hundreds of hours and no required path. The win is freedom, and the risk is drift without structure, where the story can dissolve into tourism. AI can populate the emptiness with contextual lore, side quests, ambient dialogue and world events generated from how the player explores.
Procedural narratives
Here nothing is pre-written. Algorithms and AI systems generate narrative fragments, traits and events on the fly, and the player's actions decide how they combine. AI Dungeon is the example. Variety is effectively infinite, and the risk is incoherence. The video calls this a perfect AI fit: language models can generate seeds, encounters and events while anchoring them to character goals and world themes.
Dialogue-driven narratives
Conversation is the gameplay. The player picks a tone, such as diplomatic, aggressive or curious, and the world answers in kind. In Mass Effect, the dialogue wheel shapes relationships, loyalty and even romances. The cost is that every branch needs writing, and too many choices can exhaust the player. AI writes variations, shifts emotional tone, adapts NPC personalities and reacts to what the player actually said, cutting the workload that made this structure expensive.
Environmental narratives
There are no dialogue trees or cutscenes. The story lives in a bloodstain, a broken photo or graffiti, and the player is the detective. Gone Home tells one family's history with no combat. The risk is that players walk past the story. AI acts as set dresser, generating notes, logs, graffiti and item descriptions and placing them where the player's exploration pattern will find them.
Hybrid narratives and the AI fit
Hybrid games layer everything: a linear spine, branching missions, environmental depth and procedural events. Red Dead Redemption 2 is the example, with a cost of enormous resources and difficulty holding tone together. Here AI acts as glue, providing dynamic dialogue for linear missions, emerging events for the open world, procedural side quests and adaptive lore.
The video's comparison table is described as a menu, not a ranking. Each type trades something: branching trades writing for agency, open-ended trades structure for freedom, procedural trades coherence for infinity, and dialogue trades brevity for intimacy. In the AI suitability matrix, procedural and dialogue-driven designs are a very high fit, branching, open-ended and hybrid are high, and tightly authored linear stories need the most careful prompting. In practice, developers use LLMs to draft branches and variants, RAG to keep generated lore consistent, procedural systems in Unity and Unreal to evolve quests in real time, and dialogue engines to rewrite lines to match personality.
Key takeaways
- Six delivery systems exist: branching, open-ended, procedural, dialogue-driven, environmental and hybrid.
- Each type has a price, so know it before choosing.
- Branching costs writing, and AI pays that cost with variation.
- Procedural and dialogue-driven designs are the best fit for language models.
- In hybrid games, AI holds the layers together.
- Choose the narrative structure first, then decide what the model is allowed to do inside it.
Try it yourself
Do the chapter's own exercise: sketch a branching flowchart for an abandoned house you find in a storm. Give it three decision points and two distinct endings, and make sure every choice visibly causes its consequence. Then consider which parts of the flowchart a language model could draft as variants and which you would keep authored.
Full transcript(auto-generated, with timestamps)
[0:00]Generative AI for game development. Types of interactive narratives. Six ways players live inside a story and which one's generative AI supercharges. Hello, this is Liam in forbear. When people hear interactive narrative, they picture dialogue choices. Choices are one delivery system. This chapter maps six and shows which one's generative AI supercharges. Six terms, then we're off. Branching narrative choices split the story into paths. Open-ended, the story emerges from exploration. Procedural systems generate story on the fly. Dialogue-driven conversation is the mechanic. Environmental, the world tells the story. Hybrid, all of the above, layered. Last chapter was the blueprint. How stories are structured. This chapter is the experience. How the player meets
[0:44]The story. Moment to moment. Choice by choice. Same bones, different question. Six delivery systems. Branching. Your choices fork the plot. Open-ended. The world is the plot. Procedural. The plot writes itself. Dialogue driven talk is action. Environmental, the set is the script. Hybrid, everything at once. Why learn all six? Because the structure you choose decides your workload, your players freedom, and critically where generative AI can actually help match the tool to the story. Branching at key moments, the player chooses, and the story splits. Some branches rejoin the main road. Others drive off a cliff into a completely different ending. The Walking Dead by Telltale. Every lie,
[1:24]Every mercy, every abandonment is tracked and thrown back at you episodes later. Best for narrative RPGs and moral dilemmas. The price writing multiplies with every fork. This is where AI earns its keep. It generates the variations, alternate dialogue, consequences, scene rewrites, and fills the gaps between branches so the story stays coherent no matter which fork the player took. Open-ended, no chapters, no missions, just a world. The story arrives in fragments. A rumor, a ruin, a stranger's warning. You assemble it by living in it. Skyrim, hundreds of hours, no required path. Best for open worlds and explorers. The win is freedom. The risk is drift without structure. The story
[2:05]Can dissolve into tourism. AI populates the emptiness. Contextual lore, side quests, ambient dialogue, world events, all generated from how the player explores. The world feels reactive even where nothing was handwritten. Procedural. Nothing is pre-written. Algorithms and AI systems generate narrative fragments, traits, and events on the fly, and your actions decide how they combine. AI dungeon type anything, and the story continues. No branches because nothing was written. Best for sandboxes and improvisers. Infinite variety. The risk is incoherence. The perfect AI fit. Language models generate the seeds, encounters, and events while anchoring them to character goals, and world themes so the improvisation stays coherent. Dialogue-driven conversation is the gameplay. You pick a tone,
[2:53]Diplomatic, aggressive, curious, and the world answers in kind, shifting attitudes, and redirecting the plot. Mass Effect, the dialogue wheel shapes shepherd, diplomat, or soldier in relationships, loyalty, even romances follow, best for character RPGs. The cost, every branch needs writing, and too many choices exhaust the player. AI's home turf. It writes the variations, shifts emotional tone, adapts NPC personalities, and reacts to what the player actually said. Cutting the writing workload that used to make this structure so expensive. Environmental, no dialogue trees, no cutscenes. The story is in the bloodstain, the broken photo, the graffiti, and you're the detective. Miss the clue, miss the story. Gone home, one
[3:36]House, one family's history, zero combat. Learned entirely by snooping. Best for mystery and horror. The risk players walk right past the story. AI becomes the set dresser. It generates the notes, logs, graffiti, and item descriptions and places them where your exploration pattern will actually find them. Hybrid all of it layered. A linear spine for the main story. Branching missions, environmental depth in every room, and procedural events filling the gaps between. Red Dead Redemption 2, a linear epic wrapped in dialogue missions, random encounters, and environments that tell their own stories. Best for AAA worlds. The cost, enormous resources, and tone is hard to hold together. Here, AI is the glue.
[4:18]Dynamic dialogue for linear missions, emerging events for the open world, procedural side quests, adaptive lore, holding the layers together so the seams don't show. Side by side, branching trades, writing for agency, open-ended trades, structure for freedom, procedural trades, coherence for infinity, dialogue trades, brevity for intimacy. The table isn't a ranking, it's a menu. In the AI suitability matrix, procedural and dialogue driven are a very high fit. That's what language models do. Branching, open-ended, and hybrid are high. Tightly authored linear needs the most careful prompting. How developers actually use it. LLM's draft branches and varants. RAG keeps generated lore consistent. Procedural systems in Unity and Unreal
[5:01]Evolve quests in real time. Dialogue engines rewrite lines to match personality. The key decision. Your narrative type determines how the AI generates, adapts, and delivers. Choose the structure first, then decide what the model is allowed to do inside it. Don't chase complexity for its own sake. Choose the structure that serves your story, your mechanics, and your players, and let AI do what that structure needs most. The split in six lines. One, six delivery systems, branching, open-ended, procedural, dialogue-driven, environmental, hybrid. Two, each trade something, know your price. Three, branching price is writing. AI pays it with variance. Four, procedural and dialogue are AI's home turf. Five, in
[5:41]Hybrid games, AI is the glue between layers. Six, choose the structure first, then unleash the model inside it. Your turn, the chapter's own exercise. Sketch a branching flowchart for the abandoned house. You find a mysterious house in a storm. Three decision points, two distinct endings, every choice visibly causing its consequence. Types of interactive narratives at Nick Bear
More from Generative AI for Game Development — Shesh Narayan Gupta
3:3107 · AI Tools and Technologies for Storytelling | Generative AI for Game Development
4:3305 · Understanding Narrative Structures | Generative AI for Game Development
5:0108 · Ethical and Practical Considerations | Generative AI for Game Development
6:3304 · What Is Storytelling? | Generative AI for Game Development
4:0609 · Creating Branching Story Frameworks | Generative AI for Game Development
3:15