Richard Ng
Artificial intelligence systems perpetuate harmful misrepresentations of Indigenous communities through biased training datasets that exclude Indigenous perspectives, fail to recognize knowledge holder and intellectual property rights, and struggle to accurately represent Indigenous features and cultural practices. This necessitates a paradigm shift toward community-driven AI governance. This research employs Indigenous science fiction as a methodological framework to envision AI’s impact on Indigenous futures and develop critical responses. Indigenous Science fiction challenges Western centric narratives by drawing from autochthonous knowledge systems to imagine alternative futurisms. Through The Last Hawaiian, a speculative story reinterpreting the moolelo of Poliahu and Pele depicting a dystopian future where superintelligence threatens to eliminate Kanaka Maoli and their ecological knowledge, this work illustrates the dangers of settler colonial technologies. Building on these speculative imaginings, I propose KIA’I (guardian in ‘Olelo Hawai’i), a community driven AI data governance framework protecting Traditional Ecological Knowledge through Indigenous Protocols. KIA’I empowers communities to validate AI-generated data for accuracy, cultural sensitivity, and ethical representation, ensuring Indigenous data sovereignty. The system’s architecture reflects the Hawaiian principle of lokahi (unity, harmony, balance), integrating human expertise with technological tools. Central to KIA’I is the Pewa Data Trust, utilizing blockchain technology for secure, community governed data management. This trust ensures Indigenous control over knowledge preservation while potentially unlocking economic opportunities. The model advocates for transformative human-AI relationships serving Indigenous interests, grounded in pono: correct, harmonious, and beneficial relationships with the world. KIA’I represents a pathway toward responsible, respectful, and equitable AI serving Indigenous communities.