Affirmative action for algorithms extends the logic of corrective social policy into automated systems. Where public institutions have historically allocated subsidies, scholarships, or opportunities in ways that left women and girls behind, this principle calls for algorithmic decision-making to actively rebalance that distribution, piloted on longstanding social science evidence rather than assumption.
Neutral-sounding systems built on unequal histories tend to preserve inequality, not erase it. A model trained purely to predict past outcomes will predict more of the same. Affirmative action for algorithms reframes the goal from prediction to correction, treating equitable outcomes as a legitimate design objective rather than an afterthought bolted onto a finished system.
This principle is the founding idea of the Alliance itself. The 2019 Affirmative Action for Algorithms declaration, delivered at Women in Data Science Zürich by Women at the Table, catalyzed the formation of the A+ Alliance. The full A+ Declaration sets out concrete mechanisms: Algorithmic Impact Assessments, rigorous lifecycle testing, gender-responsive procurement guidelines, and open gender-disaggregated datasets.
Related projects
A+ Declaration, Gender & AI Innovation Collective, Our Ecosystem
Related resources
A+ Advisory Board, Feminist AI Network Papers, Global Directory
