Electroshocking the Past: Art History and the Political Project of Non-Generative AI

This talk presents material from Sonja Drimmer’s current project, “Extracting the Past: How AI Is Stealing History to Rob Our Future.” Broadly, the book offers an account of the artificial intelligence industry’s reliance on historical artifacts, the labor of historians, and a perverse recounting of history itself. In this talk, she will focus on non-generative forms of machine learning—computer vision in particular—as both the driving force behind some of the AI industry’s most profitable ventures and the bedrock of the political project that this industry is enabling. She will discuss how art historians working within museums and universities have—wittingly and unwittingly—played an important role in this industry’s development, and how an inordinate focus on generative AI threatens to obscure the now-mundane uses of a technology that is being leveraged to dismantle the foundations of civil society.

Sonja Drimmer is a scholar of medieval European art with expertise in illuminated manuscripts and early print. Before joining the History of Art & Architecture at the University of Massachusetts Amherst in 2013, she received her BA from Brown University and PhD from Columbia University. Her research is largely concerned with premodern notions of authorship and authority, media theory, book history, reproduction, and the aesthetics and material culture of politics. She maintains a strong interest in historiography and in particular, how mediation, reproduction, and restoration shape the reception of objects over time: this concern extends into commentary she has written for both public and scholarly venues regarding the incursion of machine learning and artificial intelligence into the humanities and art history in particular.

Modeling Culture is generously funded by an AI Lab Seed grant, with additional support from the Humanities Council and the Princeton Humanities Initiative.