How Do Human Creators Embrace Human-AI Co-Creation? A Perspective on Human Agency of Screenwriters

Paper Title

How Do Human Creators Embrace Human-AI Co-Creation? A Perspective on Human Agency of Screenwriters

Publication Info

• Topic area: Human-AI co-creation in screenwriting, focusing on human agency.
• Keywords: Human-AI co-creation, screenwriting, human agency, generative AI, creative workflows, intentionality, forethought, self-reactiveness, self-reflectiveness, professional development.

Background and Problem

• Problem / challenge: Existing research on human-AI co-creation provides only snapshots of interaction, lacking insights into how creators develop their practices over time and exercise agency in sustained collaboration with AI.
• Significance: Understanding how screenwriters adapt to AI is critical for designing tools that support creativity, preserve human distinctiveness, and mitigate risks like skill erosion or over-reliance.
• Motivation and related work: Previous studies explored AI's role in ideation, narrative structuring, and workflow enhancement but failed to address how creators evolve their strategies, regulate AI's influence, and reflect on their practices. This study builds on Bandura’s human agency theory to investigate these dynamics.

Solution

• Proposed approach: A two-week qualitative study with 19 professional screenwriters, analyzing their co-creation practices with AI through the lens of Bandura’s human agency framework.
• Novelty:
  1. Systematic investigation of how screenwriters develop co-creation practices across four properties of human agency: intentionality, forethought, self-reactiveness, and self-reflectiveness.
  2. Identification of screenwriters’ expectations for future AI systems to better support their agency.
  3. Practical suggestions for screenwriters, tool designers, and researchers to foster sustainable human-AI co-creation.
• Procedure and key techniques:
  • Three-stage study: co-creation sessions, retrospective think-aloud protocols, and semi-structured interviews.
  • Analysis of participants’ planning, evaluation, and reflection processes.
  • Thematic analysis of 2,850 minutes of transcribed audio and annotated daily logs.

Results

• Concrete findings:
  • Screenwriters used AI for tasks like content expansion, structural refinement, and knowledge acquisition, guided by evaluative criteria such as narrative alignment, structural conformity, logical consistency, and content accuracy.
  • Immediate responses to AI outputs included adoption, adjustment, and avoidance, influencing long-term strategies.
  • Participants developed confidence, reorganized workflows, and transformed their cognitive approaches through sustained AI interaction.
  • Concerns about skill erosion and over-reliance were noted, alongside a shift in viewing AI as a partner, competitor, and emotional support.
• Advantage over baselines: Unlike prior studies, this work provides a longitudinal perspective on how screenwriters adapt and evolve their practices with AI, offering deeper insights into human development in co-creation.
• Experiments / evaluation:
  • Participants completed a five-minute screenplay over two weeks using a chat-based AI platform.
  • Data collection included daily logs, retrospective think-aloud sessions, and interviews.
  • Analysis focused on the four properties of human agency and participants’ evolving strategies and expectations.
• Limitations and future work:
  • Limited to short-form screenwriting; future studies could explore feature-length or multi-episode formats.
  • Screenplays were not staged or produced, omitting real-world production constraints.
  • Sample may lack diversity; future work should include broader demographics and creative domains.
  • Long-term observational studies and measurable constructs of agency are needed.

Summary

This study investigates how professional screenwriters exercise their agency in human-AI co-creation, using Bandura’s framework of intentionality, forethought, self-reactiveness, and self-reflectiveness. Over two weeks, 19 screenwriters demonstrated active planning, evaluation, and reflection, developing strategies to integrate AI into their workflows while addressing concerns like skill erosion and over-reliance. Participants viewed AI as a partner, competitor, and emotional support, and articulated expectations for future systems to better support their agency. The findings offer actionable insights for screenwriters, tool designers, and researchers, emphasizing the need for tools that foster human development and preserve creative distinctiveness. Future work should explore diverse creative contexts and long-term human-AI collaboration dynamics.