Appendix
Table 1: Summary of related work
| Reference | Objective | Models | Dataset | Key Findings | Research Gaps |
| Anantrasirichai & Bull, 2022) | Study AI impact on creative sectors | AI creative systems overview | Secondary literature | AI reshapes creative workflows | Limited theatre-specific focus |
| Branch & Mirowski (2024) | Explore AI in absurd theatre | Generative AI narrative models | Conceptual/theatre texts | AI creates fragmented dramatic forms | Lack of empirical performance validation |
| Branch & Mirowski, 2024) | AI in absurd theatre theory | Conceptual AI theatre models | Theoretical texts | AI enables absurdist narratives | Lack of real-world testing |
| Cake (2025) | AI script collaboration | LLM writing assistant | Script samples | Improves drafting efficiency | Weak emotional depth |
| Cardon et al. (2023) | Develop AI writing literacy | AI-assisted writing tools | Educational writing data | AI literacy is essential | Limited domain-specific theatre use |
| Chakrabarty et al. (2024) | Examine AI creativity limits | Transformer-based LLMs | Text generation datasets | AI creativity is debatable | No theatre-specific analysis |
| Dayo et al. (2023) | AI in storytelling | NLP scriptwriting tools | Media scripts | Improves writing speed/structure | Low originality concern |
| Horváth (2025) | Study AI in theatre transformation | Digital theatre systems | Theatre case studies | Hybrid theatre emerging | Lack of standard dramaturgy model |
| Hoyer & Frühmorgen (2025) | AI screenwriting process | Multi-stage LLM pipeline | Screenplay datasets | Structured script generation possible | Weak emotional modelling |
| Kabashkin et al. (2025) | AI narrative modelling | Archetype learning models | Narrative corpora | AI reproduces archetypes | Limited novelty generation |
| Kavitha (2024) | Copyright issues in AI film | Legal analysis | Policy + case review | Authorship ambiguity | No theatre-specific legal framework |
| Latif et al. (2025) | AI in theatre directing | AI staging tools | Performance experiments | Enhances directing decisions | Limited live theatre validation |
| Lo Duca & Rotelli (2026) | AI in screenwriting education | Generative AI writing tools | Film school datasets | AI supports learning | Limited professional validation |
| Mirowski et al. (2023) | AI co-writing scripts | LLM co-authoring system | Screenplay experiments | AI assists script creation | Needs human refinement |
| Palanimurugan et al. (2025) | AI and authorship change | Generative AI systems | Industry data | Authorship becomes distributed | Legal-ethical uncertainty |
| Ren (2024) | AI-generated stage plays | Generative playwright models | Stage scripts | AI produces usable scripts | Emotional depth gap |
| Tsao et al. (2025) | Review AI in creative industries | Scoping review model | Multi-industry studies | AI widely adopted | Lack of theatre focus |
| Wang (2025) | AI in micro-short drama | AI production systems | Short video drama datasets | AI reshapes narrative structure | Weak artistic evaluation |
| Xu & Xie (2026) | AI in film/TV scriptwriting | NLP + generative models | Screenplay datasets | AI supports script creation | Limited narrative originality |
| Yang et al. (2026) | Multi-agent drama generation | Co-DIRECT framework | Interactive scripts | Improves coherence & interaction | High computational complexity |
Figure 1: Proposed DSA-MAG Script Generation
Figure 2: Script Length vs Genre
Figure 3: Script Distribution
Figure 4: Accuracy Comparison
Figure 5: Precision Comparison
Figure 6: Recall Comparison
Figure 7: F1 Score Comparison