Dissertação de Mestrado

StoryXPander: An AI-Assisted Graph-Based Tool for Interactive Storytelling

Ricardo Jorge Subtil Sousa2025

Informações chave

Autores:

Ricardo Jorge Subtil Sousa (Ricardo Jorge Subtil Sousa)

Orientadores:

Carlos António Roque Martinho (Carlos António Roque Martinho)

Publicado em

14 de novembro de 2025

Resumo

In recent years, the swift advancement of Large Language Models (LLMs) has generated increasing interest in their potential to enhance creativity and narrative design within interactive digital media. This study explores the integration of LLMs as collaborative tools for creating branching narratives in video games, focusing on two primary research questions: Can these systems effectively help designers create interactive stories? And can players reliably detect AI involvement in the narratives produced? To explore these questions, we developed a custom graph-based authoring environment that allows designers to generate and expand story nodes using LLM-assisted prompts. We formulate two hypotheses: H1) LLM-assisted authoring tools can improve the efficiency and usability of interactive story creation, and H2) players will exhibit limited ability to recognize AI-generated narrative content. The evaluation involved a usability study with players that will act as designers and a player-based assessment of narrative perception. The usability results yielded a mean System Usability Scale (SUS) score of 68.3, supporting H1 and indicating that the system aids creative storytelling. The results for H2 were inconclusive, as the responses of the participants to the authorship of AI remained largely neutral, indicating uncertainty about the presence of LLM involvement. In general, the findings suggest that LLM-assisted tools can significantly enhance creative workflows in interactive storytelling. At the same time, they highlight ongoing challenges related to perception of authorship, narrative consistency, and the balance of creative control.

Detalhes da publicação

Autores da comunidade :

Orientadores desta instituição:

Domínio Científico (FOS)

electrical-engineering-electronic-engineering-information-engineering - Engenharia Eletrotécnica, Eletrónica e Informática

Idioma da publicação (código ISO)

eng - Inglês

Acesso à publicação:

Acesso Embargado

Data do fim do embargo:

22 de setembro de 2026

Nome da instituição

Instituto Superior Técnico