LLM更偏爱风格而非故事内容，且偏好稳定。
Style over Story: Measuring LLM Narrative Preferences via Structured Selection
- 用200个叙事学约束设计实验，量化模型选择偏好。
- 所有模型均优先选风格，内容元素差异大且受指令影响。
- 适合关注生成式AI创作倾向的研究者和开发者。
我们提出一种基于约束选择的实验设计，用于测量大型语言模型（LLMs）的叙事偏好。该设计为理解模型的叙事选择行为提供了可解释的视角。我们构建了一个包含200个基于叙事学的约束库，并在三种不同指令类型（基础、质量导向、创意导向）下，从六种LLMs中触发选择。结果表明，模型在所有情况下均一致地优先选择风格，而非事件、角色或场景等叙事内容元素。风格偏好在不同模型和指令类型间保持稳定，而内容元素则表现出跨模型差异和对指令的敏感性。这些发现表明，LLMs具有潜在的叙事偏好，应引导自然语言处理社区在创造性领域评估与部署模型时加以考虑。
原文摘要 · Abstract (English)
We introduce a constraint-selection-based experiment design for measuring narrative preferences of Large Language Models (LLMs). This design offers an interpretable lens on LLMs' narrative selection behavior. We developed a library of 200 narratology-grounded constraints and prompted selections from six LLMs under three different instruction types: basic, quality-focused, and creativity-focused. Findings demonstrate that models consistently prioritize Style over narrative content elements like Event, Character, and Setting. Style preferences remain stable across models and instruction types, whereas content elements show cross-model divergence and instructional sensitivity. These results suggest that LLMs have latent narrative preferences, which should inform how the NLP community evaluates and deploys models in creative domains.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。