GPT Models and Video Game Narrative: A Meta-Analysis of Language Model Studies

This study explores experimental studies and case studies focusing on the incorporation of LLMs, primarily GPT models, in video game narrative elements, thus delving into the intersection of AI, particularly NLP, and video game storytelling. In order to summarize only the most relevant and recent studies, a set of four specific selection criteria was devised in a study: publication date, relevancy to the field, type of study, and methodology. Also, to further the relevancy of this study, a set of four analysis criteria was established to aid in the summarization and categorization of each selected study's key findings and main conclusions. The first three analysis criteria showcase the current state of the art of GPT models in video game narrative elements, while the fourth and final one provides implications for future use. Additionally, in an attempt to achieve generalizability of the pros and cons of the utilization of GPT models in narrative elements across a wider range of varied video game genres, this study's analysis also takes into consideration a variety of narrative elements, from quests and dialogues to even entire storyline and worldbuilding.