process 3

Residual Agency

 

Process 03 does not deal with order (as the first two processes do). It deals with intention.

More precisely, it asks what an image-based AI produces when all technical conditions are fixed, but no longer any content-related goal exists. In this process, the AI was given very strict constraints: identical settings, identical technical parameters, precisely formulated prompts, and rigorous exclusions. Control over the process was high. At the same time, however, the AI was deprived of any form of purpose. It was instructed to fulfill no motif, make no statement, convey no idea, and produce no “good image.”

 

The prompts did not define what the image should be, but only under which conditions it was allowed to come into being. Everything that suggested design, expression, or meaning was excluded via negative prompts. As a result, the AI does not work toward a goal. It produces outputs without it being clear what these outputs stand for or whom they serve. The three images of this process show no progression and no escalation. They appear serial, functional, and matter-of-fact—like outputs of a system that operates correctly but no longer pursues an intention.

 

What is important here: these images are not failure images. They do not show breakdown or malfunction. They are what remains when a system operates without anyone demanding meaning from it.

Process 03 shows that an image-based AI continues to generate images even when intention, statement, and authorship are withdrawn. What becomes visible are not ideas, but residual states of a process.

 

In the sense of Donna Haraway, this is about shared agency. The images belong neither clearly to the machine nor clearly to the person who set up the process. They emerge in between.

 

(Process 03, Iterations 1–3)

 

Created with Stable Diffusion (A1111), GPU compute via RunPod.

 

January 2026