AI as pharmakon

Abstract In struggles for Indigenous sovereignty in Taiwan, artificial intelligence has emerged as a pharmakon—something that is both remedy and poison. This is evident in Taiwan's Trustworthy AI Dialogue Engine (TAIDE), a project that uses large language models to revitalize endangered Indigenous languages. Here, we can see the profoundly contradictory effects of AI as it enables cultural and political revitalization, thus contributing to Indigenous futurity, while intensifying the environmental degradation that forecloses that futurity. Expanding anthropological notions of authorship to encompass material world‐making, I draw on feminist and decolonial critiques to argue that debates on AI and authorship ought to encompass not only AI's textual production but also the way it underwrites extractive infrastructures and fragile futures. By holding these forces in the same analytical frame, this commentary challenges anthropologists to reckon with how AI simultaneously advances and undermines Indigenous sovereignty, troubling the purity of our categories.