Publications

On August 12 and 13, 2026, Rob Reich and Deirdre K. Mulligan convened workshops, hosted by Stanford’s Institute for Human-Centered AI, UC Berkeley’s School of Information, CITRIS and the Banatao Institute, UC Berkeley Center for Long-Term Cybersecurity, and UC Berkeley Risk & Security Lab. The purpose was to facilitate robust and diverse stakeholder input into the California Department of…

Research Summary: We explore how Large Language Models (LLMs) can serve as synthetic subjects to inform strategy research. We introduce a framework for designing and running simulated experiments with LLM-powered agents. We argue that this approach is useful for rapid, low-cost prototyping of human experiments and for generating novel hypotheses. We apply the framework to the exploration–…

The Project of AI is a world-building endeavor, wherein those who fund and develop AI systems both operate through and seek to sustain networks of power and wealth. As they expand their access to resources and configure our sociotechnical conditions, they benefit from the ways in which a suite of decoys animate scholars, critics, policymakers, journalists, and the public into co-constructing…

Government Use and Procurement of AI: Perspectives from the Biden Administration examines the critical challenges and opportunities of the federal government’s use of artificial intelligence (AI) as a mission-enabling and rights-protecting tool. Mulligan and Beech distill lessons learned from developing the Biden Administration's Office of Management and Budget (OMB) AI memos, a bellwether moment…

The Life and Death of Direct File is a leadership case study that documents the launch of the IRS's first free, government-run tax-filing service. Drawing on interviews with 15 senior government officials, the report traces the decisions and practices that changed the status quo: enabling 400,000 Americans to file their taxes for free—until the Department of Government Efficiency shut it…

Algorithmic predictions are increasingly used to inform the allocation of scarce resources. The promise of these methods is that, through machine learning, they can better identify the people who would benefit most from interventions. Recently, however, several works have called this assumption into question by demonstrating the existence of settings where simple, unit-level allocation strategies…

The American People’s Research is a data and analysis project that maps federal research and development (R&D) spending across agencies and missions, documenting the public investment that has powered the internet, modern pharmaceuticals, AI, and more. Drawing on decades of federal budget data, the work makes the empirical case for why publicly supported R&D is essential to America’s…

Differentially private mechanisms are increasingly used to publish tables of counts, where each entry represents the number of individuals belonging to a particular category. A distribution of counts summarizes the information in the count column, unlinking counts from categories. This object is useful for answering a class of research questions, but it is subject to statistical biases when…

Best Practices in Public-Sector AI Governance: A Practitioner's Playbook evaluates the risks and benefits of governmental adoption of artificial intelligence (AI). The paper offers a comprehensive analysis of public-sector AI governance, from its systems and frameworks to approaches and policies, as well as a shifting federal emphasis on the speed of adoption and on reducing regulatory…

The Federal Data Field Guide is a free resource that explains how data collection works across the federal government. The guide addresses a gap: many people want to use federal data but lack basic literacy about the government's collection methodologies, regulatory contexts, and data categories. In this resource, animals serve as conceptual frameworks for the various types of federal data…

Although information visualizations are widely used to communicate data-driven insights, the role of text in these visualizations remains understudied. Titles, captions, annotations, and other text components are pervasive and influential in real-world visualization designs. Despite the prevalence of text, the visualization field has limited empirical evidence, theoretical structure, and design…

This systematic review synthesizes 54 peer-reviewed studies published between 2019 and 2025 that examine how artificial intelligence (AI) and extended reality (XR) technologies are integrated to support adaptive and personalized learning. The studies were analyzed across multiple dimensions, including learning contexts, AI applications, adaptive input parameters, software and hardware used, and…

Large language model (LLM) applications such as agents and domain-specific reasoning increasingly rely on context adaptation -- modifying inputs with instructions, strategies, or evidence, rather than weight updates. Prior approaches improve usability but often suffer from brevity bias, which drops domain insights for concise summaries, and from context collapse, where iterative rewriting erodes…

Nutrition security, an emerging concept distinct from and complementary to food security, has gained increasing interest as a focus of efforts to reduce hunger, improve nutrition, and prevent diet-related health conditions. Yet, unlike food security, which has well-established measures, nutrition security lacks standardized measures for assessment. This gap hinders the ability to evaluate…

Data visualization is a core part of statistical practice and is ubiquitous in many fields. Although there are numerous books on data visualization, instructors in statistics and data science may be unsure how to teach data visualization, because it is such a broad discipline. To give guidance on teaching data visualization from a statistical perspective, we make two contributions. First, we…

Expertise in cognitively and motorically demanding tasks, such as indoor climbing and bouldering, is often associated with enhanced planning abilities, yet the specific relationship between cognitive and motor planning in such tasks remains underexplored. This study investigates how expertise influences route planning in bouldering, with a focus on the impact of problem difficulty. We asked…

Digital loans, which provide short-term, high-interest credit via mobile phones, have exploded in popularity across low- and middle-income countries. This paper reports the results of a randomized evaluation of a digital loan product in Nigeria. Being randomly approved for a loan (among those who otherwise would have been denied) substantially increases subjective well-being after 3 months, but…

Language shows up everywhere. It's in the digital content we circulate online, and it's in our conversations with each other. It's also in the training data and generations of language models, which are increasingly integrated into our everyday lives. Language is powerful because it embeds social identities and beliefs: it expresses who we are, and shapes our understanding of each other. Thus,…

Abortion access in the United States has been in a state of rapid change and increasing restriction since the Dobbs v Jackson Women’s Health Organization decision from the US Supreme Court in June 2022. With further constraints on access to abortion since Dobbs, the internet and online communities are playing an increasingly important role in people’s abortion trajectories. There is a need for a…