Poster
in
Affinity Event: Queer in AI

Embracing Queer and Crip Complexity in Machine Learning: Dirty Resilience and Sweaty AI

Gopinaath Kannabiran · Sacha Knox

Keywords: Queer Ecology Emplaced Ethics of Care Disability Studies Sweaty Concepts Crip Kinship Queer Futurity Non-Binary Neural Networks Sweaty AI Adaptive AI Systems Queer Trouble Queer Theory Agency in AI Diversity in AI Queer Attachments Queer Participatory AI Dirty Resilience Intersectionality Historically Disenfranchised and Socio-Politically Marginalised (HDSM) Communities Impairment Phenomenology Queer Ecologies Machine Learning Bias Mitigation Ethical AI

Project Page [ OpenReview]

Abstract

Under the theme of Queer in AI, which centres ‘queer trouble’, this paper introduces and explores two interconnected concepts: 'Dirty Resilience' and 'Sweaty AI', both aimed at addressing these challenges and developing more equitable and effective AI systems. Through the proposed praxis, we advocate for engaging the diverse needs and experiences of people across different cultural contexts. By advocating for Dirty Resilience in our data practices and algorithms, and by developing Sweaty AI systems that grapple with the complexities of intersectional identities and experiences, we challenge the field to move beyond binary thinking and the limitations of sterile efficiency.

Chat is not available.