The video assembles experiments with audio and visual machine learning models and addresses the creative impulse– especially music making– in the context of polycrisis. Two quotations in script, scroll across surreal, generative imagery, underscored by music produced using large language and local models. The quotations– the first from Bertolt Brecht, and the second an anecdote about the musicians on the sinking Titanic– talk about music, song and catastrophe, and they highlight perceived roles of art making— reflecting truth and reality, facilitating dignity in the face of disaster, offering an escape through counternarrative. The quotations address what we do when we make music or exercise our creative impulse in a time of horror or despair. In some sense, these traumatic conditions are consistent throughout history, but we in the erstwhile imperial West are now experiencing a suite of crises stemming from the decay of our social institutions, and the breakdown of the post-war consensus. The precarity and scarcity that comes along with this polycrisis are in part responsible for a revival of anxieties about the place of art in society. This work resists a narrative that would reduce the creative impulse to a single modality, nor certainly to political or ethical utility. In a sense the work rejects the terms of a question that would ask what art is for, what it should do. Through machine learning models it becomes possible to manipulate a surfeit of data for aesthetic ends. Large Language Models (LLMs) work from user prompts to produce their results (audio, video, image). For instance, one might request a painted picture of a sunflower in a vista of rolling hills in the style of Van Gogh in order to produce a familiar art-historical aesthetic. For the purpose of my experiments, I have entirely avoided this style of generation, focusing instead on how these models might be productively misused. I have prompted the LLMs indirectly, with cell phone pictures, screencaps, excerpts of found poetry, descriptions of my dreams, field recordings, and my singing voice. The audio and visual material presented here emerge from these less intentional input modes. The visual outputs of these models are often uncanny or surrealist and bring to mind a wretched unconscious of our data-scraped desire, an automatism through data wielding. Endcore, as the title implies, raises questions about the use of such tools given their connection to rentier data enclosure, environmental exploitation, and tech oligarchy, while placing pressure on questions of authorship. The piece will be sold with a local model trained on my singing voice, handmade instructions for its use, and a contract outlining terms for the exploitation of my voice for other uses (speaking, singing, music production, etc.)

ABOUT THE ARTIST Ben McCarthy’s practice plays out a dichotomy between embodied and intellectual pleasure. He is drawn to the allure of sonic texture – the natural and synthetic sounds that attune and disorganize one’s perception. Through experiments in signal processing and emerging technology he seeks to arrange a sensual present that invites attention. McCarthy works from the intuition that one’s experience of sound and voice is dense with personal and collective association. With sound, text, and documentary he thinks through the social and economic conditions that produce the listening subject. McCarthy won the Dora award for Outstanding Sound Design and Composition in Independent Theatre in 2019, while his sound design and installations have shown at the Venice Biennale, Amaze Berlin, Vector Festival, Mayworks and NAISA North. He teaches courses on labour history and the connection between art and labour at George Brown College. With Cale Weir he co-runs mainstream, an experimental music event in Toronto’s downtown.