Hybrid LLM framework balances authorial control and player agency
Original title: Towards Self-Evolving Narrative Systems: Real-Time Adaptation and Long-Term Optimization with LLMs
Self-evolving optimization refines prompts via in-context learning to sustain coherence across extended interactions.
The hybrid narrative system combines pre-authored content with real-time LLM generation using semantic similarity alignment. A feedback-driven framework eliminates manual prompt engineering. Case study and participant-based evaluation show improved narrative coherence while preserving player agency.
- Semantic similarity aligns pre-authored and generated content.
- In-context learning iteratively refines prompts.
- Enhances coherence while preserving agency.
- Aligns with authorial intent and player behavior.
From the original article
1 min read at arxiv.orgTraditional interactive narrative systems offer strong authorial control and structural coherence but often restrict player agency. In contrast, LLM-driven narratives enable open-ended generation and greater flexibility, yet frequently suffer from inconsistency and diminished narrative control.
By Ruisheng Sun · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.