Errors in administrative processes are costly and burdensome for clients but are understudied. Using U.S. Unemployment Insurance data, this study finds that while automation improves accuracy in simpler programs, it can increase errors in more complex ones.
Through our research understanding the government digital service field and what workers in this field need, we want to help strengthen those existing roles and establish more pathways for promotion and career support, as well as help other teams recognize the value of these skills and create new roles.
This paper explores how legacy procurement processes in U.S. cities shape the acquisition and governance of AI tools, based on interviews with local government employees.
The Better Government Lab at the McCourt School of Public Policy at Georgetown University has developed a new scale for measuring the experience of burden when accessing public benefits. They offer both a three-item scale and a single-item scale, which can be utilized for any public benefit program. The shorter scales provide a less burdensome way to measure by requiring less information from users.
What exactly are the differences between generative AI, large language models, and foundation models? This post aims to clarify what each of these three terms mean, how they overlap, and how they differ.
Center for Security and Emerging Technology (CSET)
This paper argues that a human rights framework could help orient the research on artificial intelligence away from machines and the risks of their biases, and towards humans and the risks to their rights, helping to center the conversation around who is harmed, what harms they face, and how those harms may be mitigated.
This study examines how bureaucratic interactions differ among public assistance programs—WIC, SNAP, and Medicaid—highlighting variations in participant experiences and the psychological costs associated with each.
This resource provides examples and practical guides that explain how to use existing regulations and data sharing agreements to transfer client information or eligibility status between benefit programs.
Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender. This academic study presents an approach to evaluate bias present in automated facial analysis algorithms and datasets.
This article analyzes the strategic use of public policy as a tool for reshaping public opinion. Though progressive revisionists in the 1990s argued that reforming welfare could produce a public more willing to invest in anti-poverty efforts, welfare reform in the 1990s did little to shift public opinion. This study investigates the general conditions under which mass feedback effects should be viewed as more or less likely.
