������Ƶ

Check out M.J. student Sam McVeety’s article “” that was included in Seattle University School of Law’s most recent issue of the .

The abstract for the paper is included below:

The rapid proliferation of Large Language Models (LLMs) like ChatGPT and Google's Bard represents a new variation on an old theme in Federal Indian law: the imposition of a monolithic worldview with a false veneer of universality. From Johnson v. M'Intosh onwards, federal Indian law is characterized by non-Native actors creating rigid definitions of what it is and how to be Indigenous in the area now called the United States and reifying those definitions through legal precedent and state violence.

Having been tuned to maximize apparent credibility, LLMs similarly co-create reality, by offering ostensibly objective statements on reality. Just as legal precedents and state violence in federal Indian law creates self-reinforcing feedback loops regarding culture and societal fitness, so too do LLMs engender a self-reinforcing, colonially inflected worldview.

In this Article, I examine the connections between Federal Indian law, LLMs, Indigenous Sovereignty Data (IDSov), and the patterns that reappear across these domains. Current data gathering practices stand to amplify harm to Native interests, and Native Nations currently face difficult choices between perpetuating their own erasure or accepting gross misrepresentations. Although there are severe shortcomings in existing legal protections, there are potential legal and technical solutions to deploy that are rooted in tribal sovereignty.