HOW ALGORITHMS UNDERSTAND CHARACTER: THE CULTURAL MEDIATION OF AI MATERIAL MODELS IN THEATRICAL COSTUME DESIGN

Recent integration of artificial intelligence (AI) into theatrical costume design has prompted new forms of dramaturgical practice, yet scholarly attention remains focused on automation and efficiency, overlooking how algorithms mediate the semiotic construction of character through material classification. Addressing this gap, this study investigates how AI models translate abstract temperament traits, such as melancholy, purity, or loyalty, into quantifiable textile parameters. Through qualitative case analysis of three AI-informed productions (Berliner Ensemble’s Hamlet, RSC’s Ophelia workshop, and National Theatre of China’s The Orphan of Zhao), supported by designer interviews and technical documentation, we examine the cultural logics embedded in training data and their performative consequences. Findings reveal that AI systems reconfigure character not through direct emotional interpretation but via statistical correlations between visual archives and material properties. These algorithmic translations often reinforce culturally specific conventions while risking reductive semiotics. Crucially, designers engage in negotiated co-authorship, critically adapting outputs to align with dramaturgical intent. The study demonstrates that AI functions as an epistemic mediator between scientific models of material behavior and performance culture, necessitating what we term “algorithmic dramaturgical literacy” among theatre practitioners.