Abstracts

Narrative Instruments: AI-Based Playable Media for Storytelling

by Max Kreminski

Creativity has been proclaimed as a grand challenge for research in both artificial intelligence (in the form of computational creativity) and human-computer interaction (in the form of creativity support tools). Storytelling is an exemplary creative domain: stories are complex creative artifacts in which many different facets—including theme, plot, character, and narration—must all be brought into alignment for the story as a whole to succeed. Consequently, the development of intelligent narrative technologies represents an excellent way to improve our understanding of creativity as a computational problem. In this dissertation, I discuss my work on the use of AI to provide plot-level creativity support for creative writing—particularly through the implementation of story sifters, which can interactively or autonomously identify sites of narrative potential within large corpuses of potentially narratable events. By crafting human-playable narrative instruments (systems that can be played to produce narrative, much like musical instruments can be played to produce music) based on story sifting technologies, I illustrate how AI can help players refine vague high-level plot ideas into coherent narrative throughlines—resulting in a new form of playful AI-supported co-creative writing, with design implications for AI-based creativity support tools in a wide variety of creative domains.