# Machine vision situations: Tracing distributed agency

## TL;DR

This paper introduces a method to study how machine vision technologies contribute to actions in creative works, highlighting complex interactions between humans and non-humans.

## Contribution

The paper introduces the concept of the 'machine vision situation' as a new analytical unit for tracing distributed agency.

## Key findings

- The method reveals key aspects of distributed agency in creative works.
- It highlights messy entanglements often overlooked in agential assemblage analyses.
- The approach is method-agnostic and applicable to both qualitative and quantitative studies.

## Abstract

This article proposes a new method for tracing and examining agency in heterogeneous assemblages, focusing on the role of machine vision technologies in creative works. We introduce the concept of the “machine vision situation”, defined as the moment in which machine vision technologies come into play and make a difference to the course of events. By taking situations as the unit of analysis, we identify moments at which machine vision technologies take part in actions without reducing them to either tools or protagonists, instead allowing for more complex agential entanglements between human and non-human actors. Grounded on an interdisciplinary theoretical framework, this article demonstrates how an analytical unit such as the machine vision situation is a valuable method for tracing distributed agency. We illustrate this through three examples by applying the method to creative works – narratives, digital games, and artworks – revealing key aspects of distributed agency and calling attention to the excess, complications, and messy entanglements that might otherwise be overlooked in analyses of agential assemblages. The machine vision situation is shown to be a method-agnostic unit of analysis that can be productively incorporated in both quantitative and qualitative studies and applied to other contexts in which human and non-human actors interact.