Research on the Application of Large Language Models in Interactive Game Storyline Generation

This study focuses on the application of Large Language Models (LLMs) in interactive game storyline generation, and systematically analyzes their performance and key challenges in the dimensions of personalized response and reasoning logic. Through literature review and case studies, the study reveals that LLMs have obvious deficiencies in long-term character memory, story logic coherence and causal reasoning, which become the main obstacles to narrative consistency and player immersion experience. Based on this, the thesis proposes several optimization paths, including memory enhancement, logic modeling, controlled generation and human-computer co-creation. The study concludes that if the narrative control and personality adaptation ability of LLM can be improved, it is expected to achieve a more immersive, intelligent and personalized AI narrative experience, providing theoretical and methodological support for generative content in game design.