The Shadow of Situationism: AI-Generated Poetry at the Avant-Garde of the Avant-Garde
Proposal Type
Individual Talk
Location
Algorithms & Imaginaries
Start Date
July 2026
End Date
July 2026
Abstract
In the contemporary literary landscape, the rise of AI-generated poetry raises questions about authorship, creativity, and the human condition that informs the composition of literature. Audience reception highlights the distaste for such “art,” dismissing it as derivative and “uncreative.” However, AI-generated art is undoubtedly co-constructed with the act of curation, splicing, and editing necessary to make a work of literature digestible. Nonetheless, internet users and avid readers tend to refuse its essence as “art.” The question becomes is this curatorial act “enough” to classify these works as poetry, and what artistic impulse seems to be informing their production? This paper situates AI-generated poetry within the lineage of “uncreative writing” as theorized by Kenneth Goldsmith, foregrounding appropriation, remix, and textual theft as aesthetic strategy. I argue that AI-generated poetry operates at the avant-garde of the avant-garde, radicalizing Situationist détournement by recontextualizing commodified language. In this sense, the Large Language Model becomes both archive and spectacle, reframing authorship as a function of dataset and interface.
Through close reading and platform analysis, I examine two works of AI-generated poetry–Brian Kim Stefans’ Beauty Face (2025) and Charles Bernstein’s Poetry Has No Future Unless it Comes to an End (2023)– and the Large Language Models that guided their production. Stefans’ long-form poem presents a kaleidoscopic image of digital culture, juxtaposing Chat-GPT generated images of infamous avant-garde poets and artists with esoteric lines regarding computational failure/glitches. Similarly, Bernstein’s collection presents poems generated through LLMs trained with his entire corpus of poetry, showcasing a product that eerily resembles a spectral double of his existing work (exposing the algorithmically generated substrate of poetic voice). These ghostly reiterations foreground questions of artistic labor, property, and genre, inviting readers to reconsider poetry’s limits under conditions of platform capitalism.
The Shadow of Situationism: AI-Generated Poetry at the Avant-Garde of the Avant-Garde
Algorithms & Imaginaries
In the contemporary literary landscape, the rise of AI-generated poetry raises questions about authorship, creativity, and the human condition that informs the composition of literature. Audience reception highlights the distaste for such “art,” dismissing it as derivative and “uncreative.” However, AI-generated art is undoubtedly co-constructed with the act of curation, splicing, and editing necessary to make a work of literature digestible. Nonetheless, internet users and avid readers tend to refuse its essence as “art.” The question becomes is this curatorial act “enough” to classify these works as poetry, and what artistic impulse seems to be informing their production? This paper situates AI-generated poetry within the lineage of “uncreative writing” as theorized by Kenneth Goldsmith, foregrounding appropriation, remix, and textual theft as aesthetic strategy. I argue that AI-generated poetry operates at the avant-garde of the avant-garde, radicalizing Situationist détournement by recontextualizing commodified language. In this sense, the Large Language Model becomes both archive and spectacle, reframing authorship as a function of dataset and interface.
Through close reading and platform analysis, I examine two works of AI-generated poetry–Brian Kim Stefans’ Beauty Face (2025) and Charles Bernstein’s Poetry Has No Future Unless it Comes to an End (2023)– and the Large Language Models that guided their production. Stefans’ long-form poem presents a kaleidoscopic image of digital culture, juxtaposing Chat-GPT generated images of infamous avant-garde poets and artists with esoteric lines regarding computational failure/glitches. Similarly, Bernstein’s collection presents poems generated through LLMs trained with his entire corpus of poetry, showcasing a product that eerily resembles a spectral double of his existing work (exposing the algorithmically generated substrate of poetic voice). These ghostly reiterations foreground questions of artistic labor, property, and genre, inviting readers to reconsider poetry’s limits under conditions of platform capitalism.
Bio
Francisco Reyes is a 4th-Year English PhD at the University of California, Irvine. His research centers digital media, race and the internet, video games, electronic literature, and the history of computing. Tangentially, he loves avant-garde film, bit-crushed music, and wasting time on the internet.