This study proposes a quantitative framework for profiling LLM dispositions as stable, model-specific regularities in output under repeated, controlled elicitation. Using a structured narrative constraint-selection task administered across six frontier models and three instruction types, we operationalize disposition through two dimensions: "consistency", measured as cross-replication selection overlap via Jaccard similarity, and "diversity", measured as dispersion across options via the inverse Simpson index. We further introduce Narrative Landscape, a PCA-based visualization that maps each model’s selection profile into a shared space for direct comparison. Results reveal a clear rigidity–exploration spectrum across model families and show that instruction types shift the geometry of selection spaces even when scalar metrics appear similar, indicating that comparable scores can mask qualitatively distinct selection topologies.

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.

Literary works present diverse and complex character behaviors, often implicit or intentionally obscured, making character analysis an inherently challenging task. This study explores LLMs’ capability to identify and interpret behaviors of artificial beings in 11 award-winning contemporary Korean science fiction short stories. Focusing on artificial beings as a distinct class of characters, rather than on conventional human characters, adds to the multi-layered complexity of analysis. We compared two LLMs, Claude 3.5 Sonnet and GPT-4o, with human experts using a custom eight-label system and a unique agreement metric developed to capture the cognitive intricacies of literary interpretation. Human inter-annotator agreement was around 50%, confirming the subjectivity of literary comprehension. LLMs differed from humans in selected text spans but demonstrated high agreement in label assignment for correctly identified spans. LLMs notably excelled at discerning ‘actions’ as semantic units rather than isolated grammatical components. This study reaffirms literary interpretation’s multifaceted nature while expanding the boundaries of NLP, contributing to discussions about AI’s capacity to understand and interpret creative works.