Feminist epistemology for machine learning systems design

Abstract

This paper presents a series of feminist epistemological concepts as tools for developing critical, more accountable, and contextualised approaches to machine learning systems design. Namely, we suggest that the methods of situated knowledges or situating, figurations or figuring, diffraction or diffracting, and critical fabulation or speculation can be productively actualised in the field of machine learning systems design. We also suggest that the meta-method for doing this actualisation requires not so much translation but transposition - a creative and critical adaptation to speak to machine learning contexts.