(AI)sland Ecologies: Toward New Metaphors and Models of Artificial Intelligence

Abstract

This article examines the application of ecological metaphors to socio-technical systems. This is a long and contested tradition that has often been critiqued for misapplying biological principles to the understanding of socio-technical systems. The practice of linking ecology with technology, however, is not inherently problematic. In this article, I seek to demonstrate how modern ecosystems are predicated upon dualistic ideologies that allow for the subsumption of nature into techno-capitalist value extraction. When applied to AI systems as such, the ecosystem metaphor obscures the material, spatial, and interrelational roots of AI. Ecology, however, is conceived differently in Indigenous island traditions, especially across the Pacific. Here, the world is seen as a continual emergence out of rich, diverse, and complex multispecies interactions. We may thus begin to see the parallels between islands and AI as world1 making projects. This article then explores how new formulations of AI—informed by Indigenous island ontologies—can be more inclusive of not just human creators and users but also the minerals, plants, and animals that directly or indirectly impact AI’s formation. This expansive understanding compels us to confront the extractive relations that underline AI today, but also to imagine a different model in which AI systems exist not as a monolith but as multiple heterogenous forms. This vision of AI is therefore one of biotechnical diversity, which can be nurtured and restored to introduce new forms at smaller scales, thereby addressing a fuller spectrum of moral and environmental questions.

Keywords

• island studies
• artificial intelligence
• indigenous studies
• multi-species anthropology