The Effect of Task Ambiguity on Human–AI Collaboration in Color Expression Design Based on a Standardized Character Prototype

This study investigates whether task ambiguity moderates the effectiveness of human-AI collaboration in color expression design based on a standardized character prototype. A 2 x 2 between-subjects experiment was conducted with 40 undergraduate design students, who were randomly assigned to low- or high-ambiguity tasks and to either a human-AI collaboration condition or a human-only condition. All participants worked on the same neutral gray 3D mannequin in order to control form-related variation, and the AI-assisted group used Midjourney with standardized character-reference settings. Creative outcomes were evaluated in terms of originality, aesthetics, and functionality. Task ambiguity and collaboration mode both showed significant main effects across the three evaluation dimensions, and their interaction effects were also significant. Under high ambiguity, the human-AI collaboration group showed a clearer advantage in originality and maintained relatively higher levels of aesthetics and functionality than the human-only group, even though both dimensions declined overall. These findings suggest that, in color expression tasks requiring abstract semantic translation, AI is better understood not as a substitute for full character design, but as a supportive partner that helps externalize color semantics, expand alternative exploration, and stabilize performance under more weakly structured task conditions. The study therefore provides empirical evidence for designing more adaptive human-AI workflows in color-centered character design tasks.