Dependency, Data and Decolonisation: A Framework for Decolonial Thinking in Collaborative AI Research (2206.03212v2)

Abstract: This essay seeks to tie together thoughts on the political economy of academia, the inequities in access to the academic means of production and decolonial practice in data empowerment. To demonstrate this I will provide a brief analysis of the neo-colonial, extractive practices of the Western Academy, introduce concepts around decolonial AI practice and then use these to form an investigative framework. Using this framework, I present a brief case study of the AirQo project in Kampala, Uganda. The project aims to deploy a low-cost air pollution sensor network across the city, using machine learning methods to calibrate these sensors against reference instruments, providing high-quality air pollution data at a far lower cost.