Underground AI? Critical Approaches to Generative Cinema through Amateur Filmmaking

Research Background and Issues

• What problems or challenges did the authors identify?
  This paper focuses on the field of filmmaking influenced by generative artificial intelligence (AI), particularly the creative practices of amateur filmmakers in non-commercial environments. The authors argue that the emergence of AI not only transforms filmmaking technologically but also poses challenges to social control, cultural expression, and creators' autonomy. Notable issues include AI bias, threats to labor, potential homogenization of cultural expression, and implicit power structures in technology dissemination.

• Why is this issue important?
  The application of AI in filmmaking is disrupting traditional creative methods and even undermining the creative autonomy of filmmakers. During the 2023 Hollywood workers' strike, many industry professionals expressed concerns about AI replacement and automation. These issues not only impact labor rights but also threaten cultural diversity and artistic expression. Therefore, reexamining the role of AI in filmmaking and proposing a counter-mainstream, creative solution is of significant importance to the fields of technology sociology and cultural studies.

• Research Motivation and Related Work
  The authors note that amateurism has historically played a critical role in film history, particularly in underground cinema, providing a platform for creative expression that challenges mainstream norms and power structures. The intersection of generative AI with this historical context offers new possibilities for innovation in filmmaking while raising questions about cultural critique and algorithmic transparency.

Solutions

• What methods or solutions did the authors propose?
  The authors conducted an 11-week study by organizing a course experiment, inviting 20 amateur filmmakers to create a total of 15 short films using generative AI. This exploration focused on identifying key technologies and critical approaches in artistic practice. They summarized four critical paths in filmmaking:

  1. Minimization: Reducing the use of AI in filmmaking to preserve creative autonomy.
  2. Maximization: Fully utilizing AI to expose its flaws and limitations.
  3. Partitioning: Restricting AI use to specific stages or functions of filmmaking.
  4. Revitalization: Redefining AI's representational and cultural potential to create new lifeworlds.

• What is innovative about this solution?
  These paths provide new methods for integrating AI technology into filmmaking, neither fully rejecting nor accepting AI. Instead, they reexamine the role of technology in creation through critical reflection, artistic experimentation, and unique cultural expression. Additionally, the authors move beyond the automation versus augmentation debate, proposing a critical technical practice for generative AI in artistic contexts.

• What are the implementation steps and key technologies used?

1. Organize the course as an experimental environment, providing resources such as generative AI tools (e.g., Runway) and digital filmmaking manuals.
2. Encourage students to use, reject, or critically engage with generative AI to create individual or team-based short films.
3. Conduct periodic creative reflections, documenting surprises, challenges, and usage strategies during the generative process.
4. Analyze the final works through rigorous film analysis to explore critical paths and the impact of technology in each film.

Research Outcomes

• What specific outcomes were achieved?
  The study developed theoretical models for the four critical paths and analyzed them through specific film examples, revealing how generative AI enables critique and innovation through diverse filmmaking practices. Additionally, the research provides empirical data and analytical frameworks for the HCI field regarding creative generative technologies.

• What advantages does it have compared to existing solutions?
  Compared to traditional automation theories or studies on creativity support tools (CSTs), this paper emphasizes critical creative engagement and encourages the AI filmmaking community to use technology as a tool for socio-cultural critique rather than mere technical optimization.

• What are the experimental or evaluation results?
  The study demonstrated how amateur filmmakers explore the potential and limitations of generative AI through critical practices. Specific examples include A Trip in Vain, which uses AI for exaggeration and satire to critique its cultural and economic context, and LOVE IN VR, which redefines AI's role in cultural expression through human-machine interaction.

• Limitations and Future Directions
  Limitations include technical biases in generative AI tools (e.g., skin tone representation, challenges in synthesizing culturally contextual speech) and the power asymmetry caused by tool dependency. Additionally, the underground AI filmmaking culture faces inherent contradictions when combined with commercialized technological tools. Future research could explore the development of non-proprietary tools to support underground filmmaking and the role of generative AI in preserving visual narrative materials.

This study provides profound insights into the critical practices and innovations of generative AI filmmaking, while further reflecting on the complex relationship between technology and artistic culture and its future possibilities.