SCMS Panel CFP: Remix in the Age of Generative AI

            Since the publication of Walter Benjamin's 1968 essay, “The Work of Art in the Age of Mechanical Reproduction,” the media industry has witnessed a decline in originality, aura, and authenticity, especially in the realm of mass media production. This shift marked an era of mechanical reproduction, punctuated by a rise in analog media technologies, not limited to the phonograph, photography, and celluloid cinema. This era paved the way for the digital revolution, during which digital media gradually eclipsed its analog counterparts, transforming reproduction technologies into the remix culture typified by sampling, mash-ups, and fan video remakes. With the introduction of the Convolutional Neural Network, the granularity of remix culture has further shifted from shot-by-shot montages in cinema and layer-by-layer collages in animation to intricate pixel-by-pixel and vector-by-vector matrices in AI content generators. This panel asks: Has the traditional boundary between fiction, film, animation, and video games progressively blurred? In the age of artificial intelligence, has “copy and paste” emerged as a new cultural norm? How has the notion of “remix” transformed with the widespread use of editing software and AIGC (Artificial Intelligence-Generated Content)?

            The digital revolution has transformed the art of “copy and paste” from avant-garde aesthetics to an everyday reality. In 2010, Clay Shirky insightfully noted this paradigm shift, stating, “Since all data is digital (expressed as numbers), there's no such thing as a copy anymore. Every piece of data... is identical to every other version of the same piece of data” (54). Techniques once unique to the celluloid era, such as collage and montage, once symbolized by “cut and paste” actions, have now detached from their original artistic contexts to become standard computer commands. These shifts have given rise to a remix culture, anchored by “cut/copy/paste” technologies (Manovich 254; Borschke 104). This cultural transformation has been expedited by the introduction of AI-assisted image and video generation tools such as Deepfakes, Stable Diffusion, Midjourney, Runaway, Pika, Animate Diff, and the Sunno. Such generative AI technologies remix pixels and embedded vectors within the enigmatic “black box” of neural networks, further complicating our understanding of originality and authenticity.

            This panel invites essays that delve into the rapidly evolving remix culture amid seismic shifts in media technology. Inspired by Kirby Ferguson's short film series, “Everything Is a Remix”, this special issue defines “remix” in its broadest sense: the art of “cut, copy, and paste” that appropriates existing texts to create a “new” work. This encompasses mash-ups and sampling rooted in music, collage in visual arts, montage in cinematic practices, and a range of short video genres such as music videos, fan video remakes, and Deepfakes face-swap videos. We particularly welcome essays that address images, whether static or moving, generated by AI models like CLIP (Contrastive Language-Image Pre-training), GAN (Generative Adversarial Neural Net), Diffusion Model, or the Diffusion Transformer (DiT). We also encourage submissions that approach “remix” not merely as a media genre but as a conceptual tool, including essays that explore the challenges surrounding copyright, and the ethical and legal implications of appropriative art and participatory culture.

            We approach “remix” in its broadest category as “the act of recombining existing materials” and welcome contributions focusing on various (trans)media genres. Proposed topics may include, but are not limited to:

• The concept: How the definition of “remix” has evolved from VCRs and Photoshop to video editing software, deepfakes, Midjourney, and Diffusion models.
• The technique: The remix of existing styles and characters through c-ref and s-ref in Midjourney, or LoRA in Stable Diffusion.
• The genre: Whether AIGC content is an extension of existing (trans)media genres like fiction, manga, animation, film/video, and games, or a novel genre parallel to traditional media categories.
• Repetition and differences: What is the minimum consistency between the generated text and the training data? What conditions cause differences between generated content with the same prompt?
• The meta-language: The combinatory logic of the Transformer, the Diffusion Model, or the Diffusion Transformer (DiT) in comparison with montage and collage.
• Media Archeology: The unexpected intersections between post-cinematic technology and pre-cinematic aesthetics, such as collage, montage in the age of analog media, fan videos, face-swap videos, and AIGC videos.

Please submit a proposal of up to 2,500 characters (around 250 words), a title of up to 120 characters, and a brief bio of up to 500 characters (around 100 words) by August 16th to the panel organizer, Yiwen Wang, at yidingyaoluwo@outlook.com.