What's with all the Tapestries?: Intersectionality and the The Discursive Vacuum of Generative AI

Abstract

This article uses its three authors' intersectional Jewish identities to critically investigate the cultural consequences of generative artificial intelligence (GAI) services. We argue that the GAI's positivistic and denotative logics treat identity as an additive construct, which is functionally incommensurate with intersectional frameworks when users aim to generate content that pertains to multiple formulations of identity. We analyze the outputs of text-to-image generator MidJourney and the large lanague model (LLM) ChatGPT for the shadow ontologies of ethnicity, race, gender, and sexuality. We situate our argument within discourses on intersectionality driven by Black feminist scholars and queer theory critiques of social classification systems. Methodologically, we employ two related experimental techniques, rooted in diasporic Jewish epistemologies, which we call kibbitzing and futzing. We conclude with a discussion of why technological approaches to rendering intersectional identities often fail, and offer an alternative paradigm for thinking through the sociotechnical affordances of GAI.

Keywords: Generative AI, intersectionality, identity, cultural representation, race, sexuality, gender, Jewish culture