Players, creators, and AI collaborate to build and expand rich game narratives

Published August 5, 2024

By Sudha Rao , Principal Researcher Chris Brockett , Principal Researcher Bill Dolan , Partner Research Manager

Share this page

This paper was presented at the IEEE 2024 Conference on Games (opens in new tab) (IEEE CoG 2024), the leading forum on innovation in and through games.

In the fast-evolving landscape of video game development, crafting dialogues and narratives is a labor-intensive endeavor. Traditionally, creating these elements involved meticulous hand-coding, resulting in static interactions that limit player agency. However, the rise of large language models (LLMs) is introducing possibilities for richer, more dynamic narrative experiences and automating some of the more challenging aspects of game creation. Despite this advance, a key challenge with using LLMs for narrative design in games is that, without human intervention, they tend to repeat patterns.

We address this in our paper, “Player-Driven Emergence in LLM-Driven Game Narrative,” presented at IEEE CoG 2024, where we explore how LLMs can foster unique forms of creativity when players participate in the design process. Rather than replacing designers, LLMs can empower players with considerable freedom in their interactions with nonplayer characters (NPC)—characters not controlled by the players but crucial for gameplay. These interactions provide implicit feedback for designers, offering insights unattainable with traditional dialogue trees—a branching structure of player dialogue choices affecting the narrative.

Creating and designing “Dejaboom!”

To test this hypothesis, we developed a text-adventure game called “Dejaboom!” The game’s premise involves a player waking up at home with déjà vu, recalling an explosion in their village from the day before. The objective is to relive the day and prevent the disaster. Players interact with five NPCs in the village. After a set number of steps, the bomb explodes, causing the player to lose all the items they gathered but retain memories of the NPC interactions. Figure 1 illustrates the game design.

Figure 1 (game design): The figure shows the map of the village where the game takes place. It shows the various locations that the player can explore, including home, park, restaurant, library, blacksmith’s shop, and town hall. It also shows the streets connecting the various locations. In addition to these, there are also two hidden rooms, namely a lab connected to the library and a storage room connected to the blacksmith’s shop. There are several objects placed at various locations that the player can pick up and use. There is a water bucket at home, a redstone torch in the park, shears in the blacksmith’s shop, a journal in the library, a map in the townhall, and a bomb in the storage room. There are five NPCs in the game that the player can interact with. There is Chef Maria in the restaurant, Mrs. Thompson on the residential street, Mad Hatter in the park, Merlin in the lab and Moriarty in the town hall. — Figure. 1. A map of the village shows the locations, objects, and NPCs.

We built the game using TextWorld, an open-source, extensible engine for text adventure games, modifying it to include dialogue with NPCs through OpenAI’s GPT-4 model. TextWorld provided the core game logic, while GPT-4 allowed for dynamic input and output—including both game feedback and NPC responses. Figure 2 illustrates our implementation of the game. In a conventional text game, this setup would allow only a fixed set of player commands and offer a predefined set of game responses. However, the use of GPT-4 allows the game’s input and output to be dynamic.

Figure 2 (game implementation): The figure depicts the implementation of the Dejaboom game. When a player issues a text command, it is first processed by an LLM which classifies it as either an action or words. If it is an action (for example “chase the birds”), then it goes to the fixed game agent which generates a fixed game response (example “this verb is not recognizable”). This response is taken in by another instance of the LLM which generates a more palatable natural language response (example “You tried to chase the birds, but nothing happened”) which is then shown to the player as the game feedback. If the player's text command is classified as words by the LLM classifier (example “can I see your menu”), then it goes to the second instance of the LLM which generates an appropriate NPC response that gets shown to the player (example “Chef Maria: Of course! Our menu today features a delicious selection of Italian-American fusion dishes”). — Figure 2: In our implementation of the game, the user’s commands are classified by GPT-4 as actions or words. Actions are processed by the game agent, while words trigger GPT-4 to generate contextually appropriate NPC responses.

Narrative analysis and user study

Our goal was to identify narrative paths that players create and how they diverge from the designer’s original narrative. We used GPT-4 to transform player game logs into a narrative graph, where a node represents a player’s strategy at specific points and directed edges (arrows) show game progression. We compared these to a graph of the designer’s intended narrative. We defined emergent nodes as those that appear in the narrative graph of players but are not present in the original narrative graph.

When we applied this approach to a user study with 28 gamers playing Dejaboom!, we found that players often introduced new strategies and elements, indicating a high level of creative engagement. Those generating the most emergent nodes tended to enjoy games that emphasize discovery, exploration, and experimentation, suggesting that such players are ideally suited for a collaborative approach to game development.

Figure 3 (narrative graph showing emergence): The figure shows a graph with nodes and edges. There are two types of nodes (blue nodes and green nodes). The blue nodes make up the initial narrative graph intended by the game designers whereas the green nodes indicate a few examples of the emergent nodes created by players implicitly through their gameplay. There is also a single start node and a single end node. A single path from the start node to the end node indicates one possible way to stop the explosion. — Figure 3: The single circles indicate the initial narrative graph intended by the designers. The double circles denote the emergent nodes created by players, representing creative new paths.

Implications and looking ahead

Our goal is to build methods that help empower game creators to create novel NPC experiences, design new narratives, and ultimately build entire new worlds through implicit player feedback and progressive application of advanced AI technologies. This work represents a foundational step, marking the start of a new paradigm of game development in which designers, players and generative AI models can collaboratively design and evolve games. Utilizing AI models introduces a new mechanism for capturing implicit player feedback through their emergent behaviors.