The paper, titled Player-Driven Emergence in LLM-Driven Game Narrative, explores, using a text-adventure game as a testbed, how players can influence emergent behaviors in games narratives through interaction with large language models (LLMs), specifically GPT-4. The study found that players' engagement with the non-deterministic aspects of LLMs led to the discovery of new, engaging game narratives not originally part of the game design. The authors, Xiangyu Peng and team, recruited 28 gamers for their research, offering insights into player motivations and behaviors in discovery-focused gameplay.