Underground AI? Critical Approaches to Generative Cinema through Amateur Filmmaking Brett A. Halperin Human Centered Design & Engineering University of Washington Seattle, Washington, USA bhalp@uw.edu Diana Flores Ruíz Cinema & Media Studies University of Washington Seattle, Washington, USA dfruiz@uw.edu Daniela K. Rosner Human Centered Design & Engineering University of Washington Seattle, Washington, USA dkrosner@uw.edu Abstract Amateurism (e.g., hobbyist and do-it-yourself making) has long helped human-computer interaction (HCI) scholars map alterna- tives to status quo technology developments, cultures, and practices. Following the 2023 Hollywood flm worker strikes, many scholars, artists, and activists alike have called for alternative approaches to AI that reclaim the apparatus for co-creative and resistant means. Towards this end, we conduct an 11-week diary study with 20 ama- teur flmmakers of 15 AI-infused flms, investigating the emerging space of generative cinema as a critical technical practice. Our close reading of the flms and flmmakers’ refections on their processes reveal four critical approaches to negotiating AI use in flmmaking: minimization, maximization, compartmentalization, and revitaliza- tion. We discuss how these approaches suggest the potential for underground flmmaking cultures to form around AI with critical amateurs reclaiming social control over the creative possibilities. CCS Concepts • Applied computing → Arts and humanities; Media arts. Keywords AI, AI Art, Amateurism, Bias, Creativity, Critical Humanistic In- quiry, Cinema, Cinematography, Critical Technical Practice, Film, Filmmaking, Generative AI, Non-Use, Storytelling, Underground Film, Video, Visual Storytelling ACM Reference Format: Brett A. Halperin, Diana Flores Ruíz, and Daniela K. Rosner. 2025. Under- ground AI? Critical Approaches to Generative Cinema through Amateur Filmmaking. In CHI Conference on Human Factors in Computing Systems (CHI ’25), April 26–May 01, 2025, Yokohama, Japan. ACM, New York, NY, USA, 18 pages. https://doi.org/10.1145/3706598.3713342 1 Introduction American independent flmmaker Jim Jarmusch describes himself in a 2017 interview as an amateur “because the origin of the word amateur means ‘the love of a form’ and professional means ‘I do this for money”’ [84]. As a prolifc flm director, screenwriter, producer, Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for proft or commercial advantage and that copies bear this notice and the full citation on the frst page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s). CHI ’25, Yokohama, Japan © 2025 Copyright held by the owner/author(s). ACM ISBN 979-8-4007-1394-1/25/04 https://doi.org/10.1145/3706598.3713342 and editor whose many international accolades include Cannes Film Festival awards, his identifcation with “amateur” cleaves the colloquial usage of the term from designating a lack of expertise. Indeed, Patricia Zimmermann’s foundational inquiry into practices and ideologies of amateur flmmaking reveal that “amateur flm is not simply an inert designation of inferior flm practice and ideology but rather is a historical process of social control over representation” [103]. As such, this study embraces amateurism as a non-hegemonic orientation that encompasses a wide spectrum of technical skill profciency and compelling aesthetic approaches. Broadly construed, amateurism has historically opened up space for alternative forms of creative expression that connect to the ethos and context of underground cinema, where flmmakers defy mainstream norms and embrace innovation in community. With the wider availability of portable flm cameras, reversible flm stocks, and low-budget optical printing in the late 1950s on- wards, a diversity of amateur flmmakers fostered community around anti-establishment, formally innovative flms often encapsulated by the term underground cinema. Underground cinema encompasses independently made and distributed flms with a strong focus on cinematic form and personal expression rather than traditional nar- rative arcs. Given their anti-establishment disposition, underground flmmakers refute an idealized flm style by embracing novel tech- niques and incorporating emergent technologies. As Parker Tyler chronicled in 1969, the term “underground” became increasingly popular in the US and UK in the 1960s, eclipsing the institutional absorption of “avant-garde” and soliciting more collaboration than “experimental” [97]. To be clear, underground flmmakers are “am- ateurs” in their non-hegemonic and inventive orientation, which exceeds conventional categorization as novices or newcomers. Our study brings flm studies concepts of the “amateur” (incorporating practitioners from novice to expert) to bear on the under-examined possibilities and politics of generative artifcial intelligence (AI) in cinema. We fnd that amateur orientations to the production of gen- erative cinema yield the potential for underground AI flmmaking. We defne generative cinema as an emerging design space where neural networks—machine learning techniques such as difusion models, visual transformers, natural language generation/processing, and computer vision—support various phases of flm production by analyzing patterns in large datasets to statistically model and generate content (audio, visual, or text). While complex technolo- gies, automation, computer-generated imagery (CGI), and software have long played a role in flmmaking [11, 66, 68], the multimodal content of generative AI suggests potential to undermine a wider set of practices—even historically analog ones like screenwriting.