This scientific–artistic dissertation examines the role of Large Language Models (LLMs) in screenwriting. Its central research question is: How can collaboration with an LLM contribute to the creation of an artistically meaningful screenplay?
To address this, the thesis establishes a “Human–AI Writers’ Room”: a controlled experimental setting for mixed-initiative co-creation, with explicit criteria for “artistic meaningfulness.” A variety of narrative experiments were conducted to identify recurring weaknesses - such as algorithmic homogenisation, bias, and narrative drift - and to develop strategies for mitigating them.
Building on these findings, the feature screenplay Sete Rios was then written over 31 days through an orchestrated Human–AI workflow in which the human writer and the LLM rotate through distinct, role-based contributions.
Drawing on computational creativity, narrative intelligence, and the philosophy of technology, the dissertation analyzes the evolving human–LLM relationship and introduces a structured practical methodology - the Causal-Thematic Prompting Framework - for mixed-initiative writing. This framework enables writers to orchestrate an LLM as a complex, instrumented, guard-railed collaborator rather than a simple automated tool, while also answering the broader question: are LLMs tools or partners within an artistic process.
Keywords: human–AI co-creation, artistic research, screenwriting, LLMs, GenAI, mixed-initiative co-creation, meaningful screenplay, narrative causality, authorship, algorithmic homogenisation, bias, narrative drift, compensatory orchestration, Causal-Thematic Prompting Framework
- Project lead: Tobias Frühmorgen
CV:
Tobias Frühmorgen (ORCID: 0000-0001-6133-8302) is a film maker, film lecturer, and film researcher. At Lusófona University he teaches Creative Production in BA and project and thesis supervision on the masters’ programs KinoEyes. He is part of a variety of EU / Horizon 2020 funded research programs like Filmterm, future.film.education, ReSTART, C-Accelerate, Cyanotypes, Crescine, RIT and Wire. Since 2020 he is part of the curriculum, pedagogy and artistic research development of FilmEU. Tobias is supervising international film projects, coordinating a joint research project on LLMs and screenwriting, and overseeing the FilmEU Micro-Credentials academy. Tobias also works as a story analyst for screenlays, is a certified Green Consultant, is co-leading a green course at Lusófona University and is a member of the CILECT Committee for Sustainability. He is completing his scientific-artistic PhD on AI and Screenwrit-ing at Filmuniversity Babelsberg.
- Contact: tobias(at)menschenkind.com
- Website: tobiasfruehmorgen.de
- PhD Advisors: Prof. Dr. Christine Reeh-Peters, Prof. Dr. Björn Stockleben
- Scientific / artistic PhD in the discipline: Screenwriting / Dramaturgy