This research investigates the impact of integrating
context-aware large language model-based non-player charac-
ter (NPC) conversations on player engagement in role-playing
games. By utilizing the capabilities of context-aware dialogues
generated by language models, the study aims to enhance player
engagement with context-aware NPCs. Through an experimental
study (n=21), measuring the player engagement and perceived
conversation quality, the results demonstrate that context-aware
LLM-based NPC conversations have high potential in increasing
player engagement in RPGs. Design guidelines for implementing
context-aware NPC interactions are identified, while acknowledg-
ing the limitations of the study. Further research is recommended
to explore other aspects of the topic.