Extracting the Past: How the 'AI' Industry Exploits Art History & What We Can Do to Stop It

Over the last several years, universities and museums have partnered with commercial technology firms like Google, Microsoft, and Meta, who have promised that their AI products will enhance both historical research and accessibility to historical collections. These promises, however, are not supported by the reality of what computer vision—the branch of AI most relevant to the history of art—can achieve. So why have major institutions in education and the arts been so quick to take up these firms’ offers?

This talk responds to this question by providing an introduction to computer vision’s origins in military surveillance, an overview of its development under late capitalist regimes of exploitative micro-labor, and an orientation to how computer vision works. However, the main focus of this talk is not what computer vision does. Rather, I consider the culture of the AI industry, its main objectives, and the dangerous vision for the future that it promises—and whether those promises are credible or even in good faith. This vision for the future has relied on extracting history, and art history in particular, and I argue that it is our responsibility as art historians to be knowledgeable about the forms this extraction takes. I conclude with suggestions about what we can do to protect the subjects and practitioners of our discipline, as well as education in the humanities more broadly, against this incursion. I do not intend an intransigent rejection of a given technology; rather this talk articulates a challenge that is grounded in knowledge of the historical origins and corporate practices of the AI industry today.

FREE