Concerns over risks from generative artificial intelligence systems have increased significantly over the past year, driven in large part by the advent of increasingly capable large language models. But, how do AI developers attempt to control the outputs of these models? This primer outlines four commonly used techniques and explains why this objective is so challenging.
Center for Security and Emerging Technology (CSET)
This academic article develops a framework for evaluating whether and how automated decision-making welfare systems introduce new harms and burdens for claimants, focusing on an example case from Germany.
An America where no one experiences poverty is possible. Already, the U.S. has programs with the potential to make this vision a reality, including programs that provide cash assistance, like Temporary Assistance for Needy Families (TANF). The current TANF program provides very little cash assistance and is marked by stark racial disparities, but it has the potential to reduce child poverty, increase economic security, and advance racial equity. This report offers a vision for an anti-racist approach to the TANF program, with new statutory goals and policy recommendations to advance racial justice.
Little is known about how agencies are currently using AI systems, and little attention has been devoted to how agencies acquire such tools or oversee their use.
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.
This guide discusses general characteristics shared by organizations that have successfully created accessible content, and includes case studies that showcase characteristics of successful accessible content teams.
This book is an in-depth exploration of federal programs and controversial legislation demonstrating that administrative burden has long existed in policy design, preventing citizens from accessing fundamental rights. Further discussion of how policymakers can minimize administrative burden to reduce inequality, boost civic engagement, and build an efficient state.
This article examines how administrative burdens in U.S. social safety net programs have changed over the past 30 years, showing that while average burdens have declined, inequality in who faces these burdens has grown.
The ANNALS of the American Academy of Political and Social Science
The DBN’s Rules as Code Community of Practice (RaC CoP) creates a shared learning and exchange space for people working on public benefits eligibility and enrollment systems — and specifically people tackling the issue of how policy becomes software code.
Professor Don Moynihan discusses how administrative burden is an effective tool to make it difficult for people to access certain types of benefits, noting that this is particularly harmful to communities of color.
This study examines the adoption and implementation of AI chatbots in U.S. state governments, identifying key drivers, challenges, and best practices for public sector chatbot deployment.
