Elsewise: Authoring AI-Based Interactive Narrative with Possibility Space Visualization

TL;DR

This work tackles the challenge of aligning authorial intent with player experience in AI-based interactive narratives by introducing Bundled Storylines and the Elsewise authoring tool. Bundled Storyline Visualization provides 1D and 2D views of the narrative possibility space, derived from author-defined or data-derived narrative dimensions, to help authors perceive, compare, and steer emergent story trajectories. A user study (n=12) demonstrates that Elsewise improves authors' anticipation of player-experienced narratives, increases perceived control, and enhances creative exploration through rapid, data-grounded feedback. The approach offers a practical pathway toward balancing authorial intent and player agency in AI-driven INs and suggests design directions for future general AI agent tooling with possibility-space visualization.

Abstract

Interactive narrative (IN) authors craft spaces of divergent narrative possibilities for players to explore, with the player's input determining which narrative possibilities they actually experience. Generative AI can enable new forms of IN by improvisationally expanding on pre-authored content in response to open-ended player input. However, this extrapolation risks widening the gap between author-envisioned and player-experienced stories, potentially limiting the strength of plot progression and the communication of the author's narrative intent. To bridge the gap, we introduce Elsewise: an authoring tool for AI-based INs that implements a novel Bundled Storyline concept to enhance author's perception and understanding of the narrative possibility space, allowing authors to explore similarities and differences between possible playthroughs of their IN in terms of open-ended, user-configurable narrative dimensions. A user study (n=12) shows that our approach improves author anticipation of player-experienced narrative, leading to more effective control and exploration of the narrative possibility spaces.

Figures (9)

• Figure 1: Example Bundled Storylines for single narrative dimension (a) and two narrrative dimensions (b)
• Figure 2: Multiple BSV views with dimension value filter. Red dashed boxes represent selected dimension value, and the BSV at the bottom represents how other BSV views on the same canvas change reflecting the filtering operation.
• Figure 3: Multiple BSV views with storyline filter
• Figure 4: Multiple BSV views with timeline control slider
• Figure 5: Illustration of the Elsewise user interface (font enlarged for readability). Dots represent narrative states in experienced storylines. Dots of the same color are from the same storyline. Exclamation marks on dots representing a triggered rule at the narrative state. (A) Users can create BSVs by defining new narrative dimensions; (B) Users can create BSVs by prompting the system to extract a set number of data-derived narrative dimensions with concept induction on playthrough data; (C) Users can create 2D BSVs by crossing dimensions from existing BSVs; (D) A 1D BSV with timeline; (E) A 2D BSV with row and column each representing a different dimension; (F) Rule editing interface; (G) Storyworld editing interface; (H) Users can prompt the system to generate a set of simulated playthrough; (I) Users can upload playthrough data to the system; (J) When users click on a dot representing a narrative state, a panel shows up to provide details of the narrative state, and allows the user to quickly navigate to the previous and next narrative state on the same storyline; (K) Compact 1d BSV views which simply classifies all the narrative states into each categorical value of the dimension; (L) Timeline control slider.