Playful approaches to defamiliarizing generative AI

Submission Type

Paper

Start Date/Time (EDT)

19-7-2024 4:45 PM

End Date/Time (EDT)

19-7-2024 5:45 PM

Location

Algorithms & Imaginaries

Abstract

The prevalence of generative AI tools makes it increasingly important to make the underlying logics of these systems legible and accessible for critique (Hemment, Murray-Rust, et al. 2023; Lindley et al. 2020; Pötzsch 2022). In this paper, I explore the playful use of defamiliarization (Shklovsky 1965), or the slowing down of perception by breaking expectations, as a way to do this. I examine my experiences participating in two netprovs, or networked improvisations (Wittig 2021) that explored the issues surrounding generative AI: the Grand Exhibition of Prompts (Marino and Wittig 2022) and HairWars (Marino 2023). Both works encouraged participants to look beyond the usual practice of writing prompts for generative AI, and to instead think about what is happening around and behind this practice. In the Grand Exhibition of Prompts, participants were encouraged to focus on the prompt, rather than the resulting image, and consider the prompt as the product. I will discuss my experience of writing long, descriptive prompts without any thought for the output, and the surprising and defamiliarizing results of running those prompts through Midjourney. In contrast, HairWarsstarted as an exploration of “accidental” encounters with intellectual property through deliberately obfuscated references to a well-known film franchise. Inspired by the Grand Exhibition, I wrote extended narrative pieces as prompts, to see what the machine would dream. I also experimented with feeding the generated text that appeared in the images back into the machine. These playful practices, pushing the boundaries of the expected input (and output) of the generative AI systems, draw attention away from the imagined perfection of the machine-generated images, and instead focus on the process, and what is happening underneath. I argue that taking a playful, defamiliarizing approach to interacting with generative AI is a productive way to undermine and interrogate the invisible systems around us.

References

Hemment, Drew, Dave Murray-Rust, Vaishak Belle, Ruth Aylett, Matjaz Vidmar, and Frank Broz. 2023. ‘Experiential AI: A Transdisciplinary Framework for Legibility and Agency in AI’. arXiv. https://doi.org/10.48550/arXiv.2306.00635.

Hemment, Drew, Matjaz Vidmar, Daga Panas, Dave Murray-Rust, Vaishak Belle, and Aylett Ruth. 2023. ‘Agency and Legibility for Artists through Experiential AI’. arXiv. https://doi.org/10.48550/arXiv.2306.02327.

Lindley, Joseph, Haider Ali Akmal, Franziska Pilling, and Paul Coulton. 2020. ‘Researching AI Legibility through Design’. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 1–13. Honolulu HI USA: ACM. https://doi.org/10.1145/3313831.3376792.

Marino, Mark C. 2023. HairWars.

Marino, Mark C, and Rob Wittig. 2022. Grand Exhibition of Prompts.

Pötzsch, Holger. 2022. ‘Reimagining Algorithmic Governance’. In Nordic Human-Computer Interaction Conference. Aarhus, Denmark: ACM Press. https://www.academia.edu/93451929/Reimagining_Algorithmic_Governance.

Shklovsky, Victor. 1965. ‘Art as Technique’. In Russian Formalist Criticism: Four Essays, edited by Lee T Lemon and Marion J Reis, 3–24. Lincoln/London: University of Nebraska Press.

Wittig, Rob. 2021. Netprov: Networked Improvised Literature for the Classroom and Beyond. Amherst College Press. https://doi.org/10.3998/mpub.12387128.

This document is currently not available here.

Share

COinS
 
Jul 19th, 4:45 PM Jul 19th, 5:45 PM

Playful approaches to defamiliarizing generative AI

Algorithms & Imaginaries

The prevalence of generative AI tools makes it increasingly important to make the underlying logics of these systems legible and accessible for critique (Hemment, Murray-Rust, et al. 2023; Lindley et al. 2020; Pötzsch 2022). In this paper, I explore the playful use of defamiliarization (Shklovsky 1965), or the slowing down of perception by breaking expectations, as a way to do this. I examine my experiences participating in two netprovs, or networked improvisations (Wittig 2021) that explored the issues surrounding generative AI: the Grand Exhibition of Prompts (Marino and Wittig 2022) and HairWars (Marino 2023). Both works encouraged participants to look beyond the usual practice of writing prompts for generative AI, and to instead think about what is happening around and behind this practice. In the Grand Exhibition of Prompts, participants were encouraged to focus on the prompt, rather than the resulting image, and consider the prompt as the product. I will discuss my experience of writing long, descriptive prompts without any thought for the output, and the surprising and defamiliarizing results of running those prompts through Midjourney. In contrast, HairWarsstarted as an exploration of “accidental” encounters with intellectual property through deliberately obfuscated references to a well-known film franchise. Inspired by the Grand Exhibition, I wrote extended narrative pieces as prompts, to see what the machine would dream. I also experimented with feeding the generated text that appeared in the images back into the machine. These playful practices, pushing the boundaries of the expected input (and output) of the generative AI systems, draw attention away from the imagined perfection of the machine-generated images, and instead focus on the process, and what is happening underneath. I argue that taking a playful, defamiliarizing approach to interacting with generative AI is a productive way to undermine and interrogate the invisible systems around us.

References

Hemment, Drew, Dave Murray-Rust, Vaishak Belle, Ruth Aylett, Matjaz Vidmar, and Frank Broz. 2023. ‘Experiential AI: A Transdisciplinary Framework for Legibility and Agency in AI’. arXiv. https://doi.org/10.48550/arXiv.2306.00635.

Hemment, Drew, Matjaz Vidmar, Daga Panas, Dave Murray-Rust, Vaishak Belle, and Aylett Ruth. 2023. ‘Agency and Legibility for Artists through Experiential AI’. arXiv. https://doi.org/10.48550/arXiv.2306.02327.

Lindley, Joseph, Haider Ali Akmal, Franziska Pilling, and Paul Coulton. 2020. ‘Researching AI Legibility through Design’. In Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 1–13. Honolulu HI USA: ACM. https://doi.org/10.1145/3313831.3376792.

Marino, Mark C. 2023. HairWars.

Marino, Mark C, and Rob Wittig. 2022. Grand Exhibition of Prompts.

Pötzsch, Holger. 2022. ‘Reimagining Algorithmic Governance’. In Nordic Human-Computer Interaction Conference. Aarhus, Denmark: ACM Press. https://www.academia.edu/93451929/Reimagining_Algorithmic_Governance.

Shklovsky, Victor. 1965. ‘Art as Technique’. In Russian Formalist Criticism: Four Essays, edited by Lee T Lemon and Marion J Reis, 3–24. Lincoln/London: University of Nebraska Press.

Wittig, Rob. 2021. Netprov: Networked Improvised Literature for the Classroom and Beyond. Amherst College Press. https://doi.org/10.3998/mpub.12387128.