以下の研究会は終了いたしました。ご参加いただいた皆さま、ありがとうございました。
On Unexpected Effects of Co-creative Practices in Algorithmic Art and Higher Education
Date and Time:Monday, May, 25, 2026, from 4PM
Venue:Room 407, 4th floor, the 14th building, University of Tokyo - Komaba Campus
Language: English
Pre-registration:Not required (Please come directly to the venue)
Speaker:Thea Sofie Engstrøm Vejlin (Aarhus University)
***
日時:2026年5月25日(月)16:00~
場所:東京大学 駒場キャンパス14号館 407教室
言語:英語
参加登録:不要(直接会場までお越しください)
登壇者:Thea Sofie Engstrøm Vejlin (Aarhus University)
共催:文化人類学セミナー
***
【Abstract】
This article examines unexpected yet persistent effects that arise in human–machine co-creative practices, particularly in experimental engagements with generative deep learning models. One such example is the emergence of an unexpected cryptid, Loab, who appears in various images as a human-like creature with violent, gory features. Loab, discovered by artist Steph May Swanson in April 2022 while experimenting with a deep learning diffusion model, went viral and quickly raised questions about how and why such unexpected images could emerge and persist in an image modelling software.
This article uses the discovery of Loab as an entry point into a broader class of contemporary socio-technical phenomena I conceptualise as monstrous effects, unsettling established ontological and epistemological categories while disrupting and reconfiguring boundaries of meaning and order (Coeckelbergh, 2019). By analysing co-creative practices in algorithmic art and higher education, I show how these effects emerge across different dimensions of entangled sociotechnical systems, both within generative models themselves and as unexpected, sometimes even unwanted, effects in institutional systems. I further argue that such effects are generated through what I term monstrous hacking: a sociotechnical practice in which humans engage with various kinds of technical, cultural, and institutional systems against their intended use, while still acting within, rather than breaking, the rules of these systems (Schneier, 2023).
Finally, I discuss how monstrous hacking is a particularly important phenomenon to foreground in contemporary debates on AI, as engagement with the ontological and epistemological specificities of deep learning technologies, namely epistemological opacity, and the ontological incommensurability of their abstractions with human thought (Fazi, 2021). Rather than treating these phenomena as anomalies, the article shows how monstrous hacking makes visible the limits of dominant Euro-American categories of agency and control and reconfigures human-machine coexistence.