Sci-Fi Spark: A Human-AI Co-Creation System for Science Fiction Ideation

要旨

In science fiction writing, ideation demands both novelty to construct fictional worlds and consistency to maintain internal and temporal logic within those worlds. While large language models (LLMs) are increasingly adopted as co-creators, generated ideas often lack surprise and struggle to maintain consistency. Moreover, current interaction paradigms of human-AI co-creation systems fail to support the fragmented and iterative nature of science fiction ideation. To address these challenges, we introduce Sci-Fi Spark, a human-AI co-creation system to support inspiration and organization in the ideation phase. The system features an Ideation Canvas for visualizing relationships between fragmented ideas, a Novelty Generator that applies computational creativity strategies to produce novel worldbuilding inspirations, and a Consistency Generator to produce context-aware storyline suggestions. A technical evaluation and a user study with writers show that Sci-Fi Spark enhances both novelty and consistency, while enabling iterative co-ideation with LLMs.

著者
Zhaojun Jiang
Zhejiang University, Hangzhou, Zhejiang, China
Wengteng Cheang
Zhejiang University, Hangzhou, China
Xuanpei Xu
Zhejiang university, Hangzhou, Zhejiang, China
Haoyu Zuo
Imperial College London, London, United Kingdom
Chunyuan Zheng
Zhejiang University, Hangzhou, China
Liuqing Chen
Zhejiang University, Hangzhou, China

会議: CHI 2026

ACM CHI Conference on Human Factors in Computing Systems

セッション: AI and Interactive Tools for the Arts

Auditorium
7 件の発表
2026-04-16 18:00:00
2026-04-16 19:30:00