Queering AI
Predictive AI systems increasingly shape social and technological life by encoding human and nonhuman entities into fixed categories, constraining indeterminate and queer senses of futurity. This dissertation develops a more-than-human design practice of queering AI, reconfiguring predictive systems toward co-predictive, relational ways of knowing and worlding.
Drawing on autotheory as a queer methodology grounded in lived experience, the research advances three experimental engagements that recode predictive systems from within, traversing AI development from data capture through post-training interaction. The thesis articulates three queer practice (re)orientations — transmutations (toward indeterminacy), algorithmic borderlands (toward thresholds), and dis/identificatory codings (toward illegibility) — and translates them into tactics for queering AI in practice.
- The greatest harm of predictive AI is not the future it predicts incorrectly, but the futures it renders unimaginable or impossible.
- Designing co-predictive relations holds open the conditions under which alternative life/worlds can emerge. (This dissertation)
- Any digital twin that claims to represent "the self" is already a political instrument that decides which versions of the self are made legible.
- Dis/identificatory codings are viable design strategies for resisting algorithmic capture without disengaging from computational systems. (This dissertation)
- Autotheory is an adept method for AI design research because it makes positionality an instrument of inquiry rather than a bias to be eliminated. (This dissertation)
- Living between neural, cultural, linguistic, and disciplinary worlds turns ambivalence into a source of creativity and critical insight.
- Prediction requires classification, and classification requires the world to be broken down into stable, distinct categories.
- The most transformative site lies in the borderlands between actual and virtual worlds. (attributed to Gloria Anzaldúa)
- Co-performance is foundational to our understanding and crafting of AI systems, as once was the notion of function to our understanding of tools. (attributed to Elisa Giaccardi and Johan Redström)
- Queerness is not a being but a doing. (José Esteban Muñoz, Cruising Utopia, 2009)