AI Governance: Tripartite Joint Adjudication and the Triangular Supervision Model for AI-related Law Enforcement Disputes

Abstract

Driven by pervasive social prejudice and bias, this study focuses on algorithmic discrimination emerging alongside widespread artificial intelligence adoption in public governance. Beyond inherent human biases, automated systems can embed encoded bias, which poses severe threats to civil rights when AI is deployed within law enforcement. Drawing on two typical U.S. cases involving wrongful detention triggered by misidentification from facial recognition and AI weapon detection tools, this paper identifies prominent institutional defects including the algorithmic black box and insufficient multi-party checks and balances under the existing regulatory framework. To mitigate such risks, the research proposes a tripartite joint adjudication mechanism, also defined as the triangular supervision model. This framework consists of professional public representatives, government technical supervisors, and AI technical developers, forming a balanced oversight system. It breaks algorithmic opacity, clarifies accountability for AI-induced harms, and transforms passive dispute resolution into proactive governance. By comparing the model with IBM’s responsible AI governance principles, this article verifies its theoretical rationality. This research argues that relevant federal and state legislation should be introduced in the United States to formalize the mechanism, ensuring AI improves administrative efficiency without undermining due process, equal protection and fundamental human rights.