AI's gender problem: A modernist perspective
Accepted version
Peer-reviewed
Repository URI
Repository DOI
Change log
Authors
Abstract
Recent commentary on AI’s ‘gender problem’ has focused on issues of representation and training in AI-related fields. Yet the white masculine bias invested in creating AI systems at the possible expense of women, minorities, and other marginalised groups transcends the number of men or women ‘doing the math’ behind the screens. The gender inequality in machine learning is only one facet of a deeper cultural bias whose origins can be found in the relationship between the female body and technology that intensified in Western modernity. This paper illustrates how the over-representation of feminised machines in nineteenth- and twentieth-century modernism presents a crucial yet overlooked background for the feminisation of humanoid robots like Sophia or the digital domesticisation of weak AI like Siri, Alexa, and Cortana in the twenty-first century. While histories of AI rarely glance past 1945 to earlier twentieth-century developments, this paper demonstrates how the intersection of feminist bodies and machines in the form of feminised robots like Maria in Fritz Lang’s Metropolis (1927), literary characters like E.T.A. Hoffmann’s Olimpia (in Der Sandmann, 1816), visual culture (such as Hannah Hoch’s photomontages or Giannina Censi’s aerodanza in the 1920s), and Luddite fears of Maschinenmensch-as-woman betray the deeply gendered bias of modern technoscientific culture. By synthesising histories of modernism with feminist theory and current AI studies, it suggests the need to re-think periodisation in the history of AI as well as the chronological distinction between ‘modernity’ and ‘postmodernity’ hypothetically represented by the singularity. While we tend to view AI as a symbol of the post-humanist moment, an historical approach reveals that machine intelligence is an extension of modernity and modernism, suggesting that we read the novelty of Haraway’s cyborgs and the revolutionary telos of the singularity incorrectly. We need to rethink the relationship between technology and modernity as well as gender and periodisation if we are to challenge and unsettle the gender bias imbuing our engagement with technology.