Paper short abstract:

Indigenous communities and their academic partners have been using machine learning applications like Google Earth Engine to map out the past, and shape the futures of their territories. I interrogate where and how these machine learning visions align with Indigenous interests & future sovereignties

Paper long abstract:

Indigenous communities – and their academic partners – have been using machine learning to help envision the past, and shape powerful stories for the futures of their places, landscapes, and territories. Huge collections of publically available digital imagery of indigenous peoples' territories have been assembled, allowing us to see cloud-free mosaics of the surface of the earth over time. Multi-spectral images and other sensors like LiDAR allows us to see through vegetation and distinguish topographic features at an unprecedented scale. Indigenous communities have harnessed these data through hand-built algorithms and machine learning using the powerful cloud-based application Google Earth Engine to map out their territories culturally significant species like seaweeds, or to detect paleo-shorelines and archaeological sites, or to demonstrate the ongoing cumulative impacts of urban and industrial developments through showing deforestation and urbanization. All of these kinds of ‘seeing’ -- made possible with powerful computational and machine learning platforms like Google’s Cloud -- are instruments to Indigenous world and future-building. They reveal with new depth and scale what is at stake in efforts and processes for reconciliation over land and resource alienation, and provide another avenue for implementing indigenous intentions and sovereignties over ancestral places, cultural landscapes and territories. This paper reveals some of the contours of Indigenous deployment of these tools, and asks a number of questions interrogating where and how these machine learning visions align with the interests of Indigenous communities.