For the Global Majority world, the dilemma is sharp. Many languages and knowledge traditions are underrepresented in training data. But “fixing” that via ungoverned scraping turns communities into data sources, not decision-makers.

I argue that is 'extractive inclusion': being included in datasets without power over collection, use, representation, refusal, attribution, or benefit-sharing. This is a form of epistemic violence. (3/n)

#LLM #data #AI #epistemicviolence #inclusion