The paper explores how interaction with large language models (LLMs) can lead to emergent behaviors in game narratives, empowering players to influence the evolution of the game. The study is conducted using a text-adventure game called Dejaboom!, where players must solve a mystery while interacting with non-player characters (NPCs) generated by GPT-4. The researchers recruited 28 gamers and used GPT-4 to convert game logs into a node-graph representing the narrative. They found that players could discover new, interesting nodes that were not part of the original narrative but added fun and engagement. Players who created the most emergent nodes tended to enjoy games that facilitate discovery, exploration, and experimentation. The study highlights the potential for LLMs to enhance game design by providing implicit feedback and fostering a more collaborative development process among designers, players, and LLMs.The paper explores how interaction with large language models (LLMs) can lead to emergent behaviors in game narratives, empowering players to influence the evolution of the game. The study is conducted using a text-adventure game called Dejaboom!, where players must solve a mystery while interacting with non-player characters (NPCs) generated by GPT-4. The researchers recruited 28 gamers and used GPT-4 to convert game logs into a node-graph representing the narrative. They found that players could discover new, interesting nodes that were not part of the original narrative but added fun and engagement. Players who created the most emergent nodes tended to enjoy games that facilitate discovery, exploration, and experimentation. The study highlights the potential for LLMs to enhance game design by providing implicit feedback and fostering a more collaborative development process among designers, players, and LLMs.