Exploring the Application of Artificial Intelligence in Government Oversight: A Case Study of the U.S. Department of Government Efficiency (DOGE)

Abstract

In January 2025, the U.S. Department of Government Efficiency (DOGE) was established to utilize artificial intelligence to oversee the operations of the federal government and the conduct of government officials. It identified a series of issues and proposed numerous reform measures, but the reforms failed rapidly due to significant controversy. The success and failure of the DOGE's government oversight and reform efforts offer three important theoretical insights. First, artificial intelligence significantly expands the boundaries of human rationality. Through the application of AI technology, there has been a significant leap in government oversight capabilities, demonstrating immense potential in assisting humans in building high-quality governments. Second, the dilemma of human nature's alienation. The "immoral economic agent attributes" of government officials (rooted in self-interest motives, rent-seeking behavior, or dereliction of duty) lead to a decline in governance effectiveness and foster systemic resistance to technological oversight. Third, the dilemma of institutional resistance and technological limitations. The potential of AI-empowered government supervision is constrained by three obstacles: the institutional monopoly of government administrative power, the "black box" of government formed by the evasion or even resistance of the government and officials to supervision, and the technical limitations of AI itself, such as algorithmic bias and data privacy. Based on this, this paper proposes a five-dimensional practical framework of "technology empowerment, institutional embedding, improvement orientation, consensus observation, and moral cultivation" to provide an effective and feasible path for the deep integration of AI into the government supervision system.