Ecofeminist Interventions in AI Systems

Abstract: AI systems are central to today’s technological landscape. But behind these infrastructures lie political and social structures that shape how they function and who they benefit. These systems rely on the large-scale extraction of labor, resources, and data. The human work that sustains them, such as data labeling, content moderation, and data cleaning, is often outsourced, poorly compensated, and rendered invisible. At the same time, the environmental footprint of AI infrastructures, from the high energy demands of training large models to the mining of hardware materials, is significant and disproportionately impacts vulnerable regions. Yet both forms of extraction remain largely overlooked in mainstream AI research. In this talk, we examine how power operates in the design and deployment of AI systems. We analyze how current practices prioritize speed, automation, and control, often reinforcing global inequalities and ecological harm. Drawing on research in data justice, intersectional feminist studies, and political ecology, we argue that AI systems are shaped by social and political choices, not just technical ones. As an alternative, we explore participatory design and community-centered approaches as strategies for redistributing power and building technologies that serve the common good.