Please note that this event is independently organized by a WASP-HS researcher and not the WASP-HS Program Office.
Such of today’s AI development is driven by large-scale models that demand vast datasets, compute resources and invisible human labour — often reinforcing the very inequalities they claim to solve. But what might AI look like if we built it differently? Drawing from the participatory, feminist work of the Data Against Feminicide project, this talk will explore how we can shift the politics of knowledge and data production in AI development towards non-extractive approaches that centre context, collaboration and care.
Data Against Feminicide is a collaborative research and design project focused on supporting the work of civil society organisations and grassroots activists who monitor gender-related violence and feminicide — the gender-related killing of cisgender and transgender women and girls. Across the world, many activists and grassroots groups produce their own data to draw attention to this systemic, lethal violence, hold public institutions accountable, support collective action and remember lives lost.
Since 2019, our interdisciplinary team has worked with activists and civil society groups across the Americas and in Sub-Saharan Africa to co-design machine learning tools that support activists’ existing data production strategies rather than replace their labour. This work has included participatory data annotation and model evaluation with attention to the socio-spatial and ethical complexities of defining and detecting feminicide across contexts.
The talk will reflect on what this work reveals about the politics of AI, showing how choices about who is involved, how technology is developed, and to what ends directly shape its social impact. It will argue for a feminist data epistemology that moves from centralised, extractive data practices towards collaborative forms of knowledge production; from abstract, generalising models towards bespoke tools attentive to local and contextual difference; and from harmful automation towards more reflexive, caring engagements that question not only what machine learning can do but whether it should be applied at all.
Using concrete examples from the project, the talk will also explore the tensions of co-producing AI: how such tools can alleviate activist labour, where biases remain and what it means to democratise technical know-how. Ultimately, it will invite researchers and practitioners to rethink how AI can be designed with and for communities on the frontlines of social justice.
The workshop is arranged in a hybrid format on 18 March, 2025
– 12.30 to 13.20: Hybrid lecture on zoom
– 12.30 to 15.30: On-site lecture plus workshop in SOL:A129b, Helgonabacken 12, Lund, Sweden
Invited speaker: Ericka Johnson, Professor at Linköping University
13.00 – 14.00 Lecture (hybrid)
14.00 Coffee break
14.15 Workshop session
To participate is free of charge. Registration for online lecture or for both lecture and workshop on-site in Lund at ai.lu.se.
