Artistic research on Recursive Dynamics in Generative Image Models
In the rapidly evolving landscape of transformer-based artificial intelligence (AI), feedback processes within generative models like Stable Diffusion are becoming increasingly significant. The talk examines the phenomenon of “self-cannibalizing” feedback loops, where the recursive use of model outputs as subsequent inputs for the model leads to unique emergent behaviors, biases and hints at a potential model collapse. By employing feedback as a strategy in various image generation pipelines, we uncover the intricate mechanisms that drive these systems, revealing the underlying tendency that shapes their functions.
Dates of Event
19th September 2024 – 20th September 2024
Poster Talk by Ting-Chun Liu & Leon-Etienne Kühr @ TEDA’24: The Magic Machine, Through the Prism of Art and Science, University of Cambridge
Conference Description
Tacit Engagement in the Digital Age (TEDA) 2024 continues the discussion from the 2019 conference. It will take place at the Faculty of Music, University of Cambridge, across 2 days on 19th and 20th September and is a collaboration between the Faculty’s Centre for Music and Science and the AI&Society Journal (Springer). It promotes values for critical thinking, drawing on a diversity of cultural perspectives and practices from across the arts and sciences to reflect on how to shape our AI futures, overcome the belief in the magic machine, and design policies for common good. The arts and humanities are considered vital in this envisioning.
