Interactive AI-driven dramaturgy for adaptive stage narratives

Thesis Abstract

The emergence of interactive AI-driven dramaturgy promises to redefine stage narratives by enabling real-time, audience-responsive dramaturgical systems that adapt character behavior, plot progression, and visual storytelling to dynamic audience input and performance context. This study addresses the problem of rigid, linear staging in traditional theatre and the need for scalable, transparent AI mechanisms that support coherent adaptive dramaturgy while preserving artistic integrity. The aim is to develop and evaluate a computational framework that integrates machine learning, affordance-based scene planning, and rule-based dramaturgy to produce adaptive stage narratives. Specific objectives include (1) designing an architectural model that integrates natural language understanding, emotion-aware character control, and real-time scene planning; (2) implementing a prototype system that synchronizes AI-driven dramaturgy with live performers, stage machinery, and audience interaction; (3) evaluating dramaturgical coherence, audience engagement, and performative quality across multiple performance scenarios; and (4) deriving design guidelines for ethical and collaborative human-AI theatre practice. The methodology combines a mixed-methods research design anchored in computational storytelling, performance studies, and human-computer interaction. The population includes professional theatre practitioners, dramaturgs, and audience members participating in stage readings and controlled performances. A purposive sample of eight professional dramaturgs and four directors, together with two production teams, will participate in iterative co-design workshops to refine the prototype. For empirical validation, two full-length performances will be staged in a mid-sized theatre space with a total audience of approximately 260–320 per night, providing 520–640 audience data points across two runs. Data collection instruments comprise (i) system logs capturing AI-driven decisions, cue generation, and latency metrics; (ii) director and performer journals and post-show interviews; (iii) a 30-item dramaturgical coherence scale and a 5-point Likert audience engagement survey administered to attendees after each performance; (iv) thematic interview guides for qualitative insights; and (v) an expert review rubric assessing artistic quality, narrative continuity, and technical reliability. Data analysis employs a convergent mixed-methods approach. Quantitative data will be analyzed through regression analysis to assess relationships between AI decision latency, narrative coherence scores, and audience engagement; time-series analysis will examine how adaptive cues influence scene progression over performances. ANOVA will compare coherence and engagement across performance modes (static vs. adaptive). Qualitative data from interviews and journals will undergo thematic analysis to identify emergent patterns in perceived dramaturgical integrity, collaboration dynamics, and ethical considerations, with triangulation against system logs. The study will also apply the Situational Theory of Theatre Engagement to interpret audience responses and incorporate Actor-Nreative Theory to evaluate character believability under adaptive control. A conceptual model of the adaptive dramaturgy framework will be proposed, detailing inputs (audience signals, performer state), processes (narrative planning, cue generation, motion control), and outputs (dialogue, blocking modifications, lighting cues). Expected findings include (a) evidence that latency and coherence management are critical determinants of audience immersion in adaptive narratives; (b) validation that a hybrid AI architecture combining deep learning for interpretation of performer cues with rule-based dramaturgy yields stable and artistically credible adaptations; (c) insight into best practices for human-AI collaboration, including transparency of AI decisions and designer-controllability over narrative trajectories; and (d) a set of design guidelines for ethical considerations, including authorship attribution, performer autonomy, and safeguarding against narrative dissonance. The study contributes to knowledge by operationalizing a scalable, auditable framework for interactive AI dramaturgy, advancing theories of computational storytelling and performance research, and informing design of theatre technologies that balance machine-driven adaptability with artistic agency. It offers practical implications for theatre education, professional staging, and the broader field of AI-enabled narrative systems. Recommendations include extending the prototype to multilingual dramaturgy, exploring real-time audience analytics for more nuanced adaptation, and developing standardized evaluation protocols for AI-assisted theatre to inform policy and practice in creative industries. The expected conclusion asserts that carefully calibrated, transparent AI-driven dramaturgy can enhance audience engagement and narrative richness without compromising artistic integrity, provided that human dramaturgs retain authoritative oversight and clear control over dramatic trajectories.