Project Description
This research project, conducted for Abundant Intelligences, explores what responsible AI infrastructure could look like for Indigenous research communities by examining alternatives to conventional, centralized AI systems. Drawing on a review of more than 40 sources and surveys from five of Abundant Intelligences research Pods, the project evaluates infrastructure options through the lenses of sovereignty, sustainability, affordability, and technical feasibility. The findings show that most Indigenous research groups prioritize data sovereignty but face significant barriers related to cost and technical capacity, leading many to rely on university-managed systems rather than community-controlled infrastructure. The research highlights the potential of smaller AI models, refurbished hardware, cooperative cloud hosting, edge computing, and renewable energy-powered systems as viable pathways for Indigenous AI development, while emphasizing that infrastructure decisions are fundamentally governance decisions about power, control, and self-determination. Ultimately, the project concludes that there is no single infrastructure model that fits all communities; instead, Indigenous communities should be empowered to choose from a range of options that align with their values, capacities, and long-term goals, while recognizing that the next challenge extends beyond infrastructure to questions of who builds AI and whose knowledge and interests shape it.
