Examining cognitive responses to evil, neutral, and good morality in AI-driven generative character arts

Variations in individual interpretation often lead to diverse moral readings of visual character representations. This study explores how moral alignment classified as Evil, Neutral, and Good is perceived in AI-generated character designs. Participants were presented with a set of carefully curated AI-generated stimuli, and their associative responses were analyzed using Associative Concept Network Analysis (ACNA) to map emergent semantic patterns. The resulting networks indicated that Evil-aligned characters were strongly associated with notions of dominance and threat (e.g. beast, murder, revenge), Neutral characters with moral flexibility and ambiguity (e.g. protect, charisma, visionary), and Good characters with guidance and inspiration (e.g. angel, mentor, wisdom). These results highlight the pivotal role of visual signifiers facial expression, costume, and color palette in shaping moral interpretation. The consistency of responses across stimuli validates the reliability of AI-generated imagery as a source of meaningful user impressions and supports its use as a tool for ideation and moodboarding. This study contributes an empirically grounded database of moral-visual associations and offers a systematic framework for assessing cognitive and affective responses to AI-assisted character design, advancing the discourse on design creativity and semiotic evaluation of generative art.