Incorporating literary rules to guide human and ai story writing
Incipit: Axiom-Grounded Scaffolding for Human-AI Literary Creation
Computers and Society
Summary
Writing long stories with AI can be tricky because the computer doesn’t explain the ideas behind the story it creates. The authors created a tool called Incipit that adds a planning step based on common storytelling rules called axioms. This plan lets writers see and change the story’s outline and characters before the computer writes the full scenes. This approach aims to make AI storytelling more understandable and controllable, though it hasn’t yet been tested to prove it makes stories better.
What this means in practice
- •For creative writing software developers: Build writing tools that show and allow editing of story plans grounded in storytelling principles before AI generates prose.$Commercial implications: Enables sale of authoring software that offers transparent narrative planning features to writers seeking AI assistance.
- •For game narrative designers: Use literature-grounded planning layers to build interactive story outlines that can be refined before automated scene generation.
A position paper. It proposes an approach and reports no results.
Authors
Qiang Liu, Chunyi Zhao
Abstract
Large language models can produce fluent prose from short prompts, but a direct interaction gives writers little access to the assumptions that shape a long narrative. We present Incipit, an implemented research prototype that inserts an explicit planning layer between writer intent and generated prose. The layer is grounded in literary axioms, which are curated and reusable propositions about human experience and narrative craft. The prototype connects a knowledge base of 1455 axioms and 472 typed relationships to a five round direction dialogue, a retrieval and selection pipeline, and a three level blueprint covering creative premises, story beats, character arcs, and chapter outlines. Writers can inspect and edit the resulting structures before using them as context for scene generation. Additional modules support real event abstraction and five dimensional diagnostic feedback. We describe the design rationale, data flow, implementation boundaries, and a worked design example. Because no controlled user study or independently rated output study has yet been completed, we do not claim that the system improves literary quality. Instead, we outline a future preregistered comparison designed to distinguish the contribution of axiom grounding from that of hierarchical planning. The paper contributes a concrete architecture for making literary knowledge an inspectable coordination object in human AI writing.