Indigenous (Mis)Representation in Emerging LLM Research Methodologies

The American road towards the present has been one that has been predominated by injustice, massacre, and genocide — a past which is often mythologized and whitewashed for patriotism and ongoing racial discrimination. This is especially relevant regarding the American invasion of western Indigenous territories and the ongoing genocides of their peoples. This study examines this historical misrepresentation through cinematic portrayals of Indigenous Americans, and the reemerging accessibility of these portrayals through conversations with Large Language Models (LLMs) and related forms of layman’s historical research. Fifty-two progressive Western films were compiled by prompting OpenAI’s ChatGPT, Google’s Gemini and AI Studio, in addition to scraping the top Wikipedia results on Google’s search engine. These films were then analyzed on various aspects of positive and negative representation (and nonrepresentation). Through this analysis, the inadequacies of LLM’s in understanding historical and cultural ethics as they pertain to modern sensibilities is illuminated. By evaluating critical dimensions in casting practices, trope prevalence, and narrative framing in the selected films, an image of a continuing and evolving cultural genocide emerges. The findings suggest that the integration of LLMs in research practices only exacerbates the spread of misinformation, undermining efforts by Indigenous academics and filmmakers to challenge reductive stereotypes. This research advocates for enhanced digital literacy and critical engagement with AI-driven tools to mitigate their detrimental effects on historical understanding and cultural representation, contributing to a broader discourse on the ethical implications of AI and its role in the preservation and dissemination of marginalized histories.