Interactive AI-assisted dramaturgy for live theatre rehearsal and performance

Thesis Abstract

This study investigates how interactive artificial intelligence (AI) systems can augment dramaturgy in live theatre rehearsals and performances, addressing the gap between traditional directive dramaturgy and data-driven creative processes in contemporary stage practice. The problem centers on limited scalability and real-time adaptability of dramaturgical decisions when productions involve large ensembles, multiple venues, and dynamic audience interactions, which often constrain iterative experimentation and rapid feedback loops. The aim is to design, implement, and evaluate a middleware-driven AI-assisted dramaturgy framework that supports dramaturgs, directors, and performers through real-time suggestion, cueing, and narrative coherence analysis without detracting from artistic agency. Specific objectives are to (1) synthesize a multimodal AI model that analyzes script, blocking, and performance data to generate dramaturgical alternatives; (2) develop an adaptive user interface that facilitates human–AI collaboration during rehearsal; (3) examine the impact of AI-assisted dramaturgy on coherence of narrative arc, actor expressivity, and rehearsal efficiency; (4) evaluate user acceptance and perceived creative autonomy among dramaturgs, directors, and actors; and (5) propose ethical and practical guidelines for deployment in professional theatre settings. The methodology adopts a mixed-methods research design combining iterative prototyping with empirical evaluation over three production cycles in a mid-sized theatre company, involving a population of 60 participants including 10 professional dramaturgs, 12 directors, and 38 actors. A purposive sample of four full-length productions, each spanning 18–24 rehearsal days, provides diverse genres (drama, contemporary, and classical adaptations). Data collection instruments comprise (i) system logs and interaction transcripts from the AI dramaturgy platform; (ii) pre- and post-rehearsal surveys measuring perceived usefulness, perceived ease of use, and creative autonomy grounded in the Technology Acceptance Model (TAM) augmented with the Self-Determination Theory; (iii) standardized performance quality rubrics assessing narrative coherence, character consistency, and stage direction alignment; (iv) senior dramaturg interviews to capture qualitative insights; and (v) audience feedback via short in-house surveys during previews. Analyses will utilize thematic analysis for qualitative interview data, regression analysis to explore relationships between AI usage and rehearsal efficiency metrics, and repeated-measures ANOVA to compare narrative coherence and actor expressivity across production cycles. A quasi-experimental component will compare AI-assisted and conventional rehearsal sessions within the same productions, controlling for director, genre, and cast variables. The analytical framework will embed N. K. Murray’s dramaturgy theory and Sutcliffe’s human–machine collaboration model, operationalized through a conceptual pipeline that maps AI suggestions to dramaturgical decisions and tracks their impact on performance outcomes. Model specification includes a Bayesian hierarchical regression to account for nested data (actress with scenes, scenes within acts, acts within productions) and the use of time-series analysis to assess changes in coherence metrics over rehearsal days. Key expected findings include (1) AI-generated dramaturgical alternatives that improve narrative coherence by providing at least two viable scene-level options per blocking sequence; (2) measurable improvements in rehearsal efficiency, evidenced by reduced time-to-block and faster resolution of continuity inconsistencies; (3) enhanced actor expressivity and alignment to character trajectories, as indicated by higher scores on the expressivity rubric and expert dramaturgic review; (4) high acceptability and perceived creative autonomy among dramaturgs and directors, moderated by perceived system transparency and control granularity; and (5) identified ethical considerations and practical constraints for on-stage AI interventions, including data privacy, authorship, and risk mitigation. The study contributes to knowledge by operationalizing an empirically validated AI dramaturgy framework that integrates multimodal data streams with human creative processes, advancing theories of human–machine collaboration in the performing arts and outlining a scalable model for professional theatres. It offers actionable guidelines for design, governance, and integration of AI-assisted dramaturgy, including user interface heuristics, decision-rights allocation, data governance, and safety protocols for live performance. The main conclusion anticipates that AI-assisted dramaturgy can augment creative decision-making without impinging on artistic agency when designed with transparent affordances, iterative human oversight, and rigorous ethical standards, with recommendations for broader field trials, platform standardization, and professional training programs for theatre practitioners.