“Maximizing Our Losses”

In this paper, I explore what becomes of visual representation in the age of machine learning, what forms of human knowledge these networks generate and reiterate, and how machine-learning-made images refer us to other images, words, and objects. As I will argue, these works repeat, with significant differences, the physical and visual signs of previous centuries. In so doing, they likewise inherit the affective expressions and referential instabilities that define the analog arts. I focus on Anna Ridler’s House of Usher series, though the claims extend to a much broader range of contemporary aesthetic experiments. Ridler’s work points us towards the nineteenth century (like so much AI thinking and making) and to Edgar Allan Poe’s short story. In turn, I trace a chain of relationships from twenty-first century computational art to nineteenth-century literature. I am interested in this particular set of words and images—in Ridler and Poe—bound together through reference and repetition, but I am also interested in what these words and images might have to teach us about the promises of indexical contact, causal reference, and visual knowledge in the twentieth and twenty-first centuries. Ridler’s work reminds us of the referential instabilities that have been “there” all along.