Data Feminism for AI

Lauren F. Klein and Catherine D’Ignazio’s paper moves data feminism into the current AI context. It does not treat fairness as a matter of tuning model metrics. Instead, it argues that AI research and practice must begin with power: who defines the problem, who supplies the data, who is categorized, who absorbs the cost of error, and who can refuse a system’s entry into their life.

The article matters because it gives AI ethics a stronger feminist method. Intersectionality, situated knowledge, co-design, and refusal of harm are not decorative value statements. They are practical principles for deciding whether an AI system should be built, how it should be evaluated, and when it should be stopped.

For FemRes, this paper pairs well with Gender Shades, Noble’s work on technology studies, and feminist data visualization. It lifts the archive’s AI-bias resources into methodology: the question is not only how to make AI more accurate, but how to make technical design accountable to power and harm.