LLM-Creativity: Measuring and Explaining Creativity in Large Language Models: A Mixed-Methods Investigation into the Effect of Cultural Persona Prompts on Business Solution Generation

This study investigates the effect of theory-driven cultural persona prompts on the creative output of Large Language Models (LLMs) in a business problem-solving context. While persona prompting is a common technique for guiding LLMs, it often lacks a structured basis, and its impact on the multidimensional nature of creativity has been underexplored. This research addresses this gap by examining how personas derived from the cultural frameworks of Hofstede and Hall influence five distinct creative criteria: Novelty, Usefulness/Feasibility, Flexibility, Elaboration, and Cultural Appropriateness/Sensitivity. An explanatory sequential mixed-methods design was employed, beginning with a large-scale quantitative experiment (N = 6,375 generated solutions) to measure the effects of 17 distinct persona conditions across a panel of five LLMs. The quantitative results informed a subsequent qualitative phase, which analyzed evaluator justifications to explain the observed statistical patterns. The findings reveal that persona effectiveness is criterion-dependent; no single persona universally enhances all aspects of creativity. A key contribution is the empirical identification of a trade-off, where personas designed to maximize novelty and flexibility did so at the expense of elaboration and perceived feasibility. Furthermore, the results indicate that genuine cultural resonance is achieved through deep procedural integration of cultural values rather than superficial acknowledgment. This research provides a robust framework for strategically directing AI creativity, demonstrating that methodical prompt design is essential for moving beyond arbitrary generation toward intentional, nuanced, and predictable creative outcomes in human-AI collaboration.