Federal Reserve Bank
of San Francisco Federal Reserve Bank
of San Francisco

Board of Governors

  • AI Adoption and Firms’ Job-Posting Behavior

    Jessica Liu, Douglas Webber

    This note uses job postings data from Lightcast and the Census Bureau’s Business Trends and Outlook Survey to investigate the relationship between AI adoption and firms’ job-posting behavior. We find that thus far, there is no evidence of a reduction in job postings for industries or firms which have higher levels of AI adoption. The overall slowdown in national job postings following the pandemic recovery does not appear to be driven (even modestly) by AI. That said, we focus on the total level of job postings in firms and industries, rather than on specific occupations which may be particularly susceptible to […]

  • Monitoring AI Adoption in the US Economy

    Jeffrey S. Allen

    This note uses three publicly available surveys with complementary target respondents to examine trends in AI adoption in the U.S. through 2025. Business survey data from the Census Bureau show that about 18 percent of firms have adopted AI as of year-end 2025. Prior to a methodological change in late 2025, the adoption rate grew by 68 percent for the year ending in September. Work-related Generative AI adoption reported by individuals in the Real-Time Population Survey stands at about 41 percent as of November, with the strongest growth occurring in the most recent quarter. Additionally, a November iteration of the Survey […]

  • Validating Large Language Model Annotations

    Anne Lundgaard Hansen

    This paper proposes a validation framework for LLM-generated measurements when reliable benchmarks are unavailable. Validity is established by testing whether an LLM can reconstruct passages from annotated labels while maintaining semantic consistency with the original text. The framework avoids circular reasoning by establishing testable prerequisite properties that must be met for a validation to be considered successful. Application to news article data demonstrates that the framework serves as a practical alternative to human benchmarking, which offers advantages in objectivity, scalability, and cost-effectiveness while identifying cases where LLMs capture economic meaning that human evaluators miss. Read the paper

  • Generative AI: What Type of Technology?

    Martin Neil Baily, David M. Byrne, Aidan T. Kane, Paul E. Soto

    Generative AI (genAI) has expanded the scope of artificial intelligence, raising both hopes and fears that it could rapidly change the structure of the economy, reduce employment in many occupations, and increase productivity. There remains great uncertainty, however, over both the time frame for the economic effects of genAI and the magnitude of its ultimate impact on the economy. In this paper, we assess genAI through the lens of productivity. Some new technologies, such as the lightbulb, temporarily raise productivity growth as adoption spreads, but the impact fades when the market saturates. Others, such as the electric dynamo, spur ongoing innovation […]

  • Quantifying Deregulation and its Economic Effects: A Large Language Model Approach

    Danilo Cascaldi-Garcia, Matteo Iacoviello

    We construct a news-based index of deregulation for the United States from 1960 through 2025 using AI to semantically classify newspaper articles. We distinguish articles discussing deregulation from those discussing increased regulation, assigning intensity scores that reflect both the centrality of deregulatory content and whether articles discuss advocacy, proposals, or enacted measures. Human validation confirms strong agreement between AI and human classifications. The deregulation index captures major reform episodes including transportation and telecommunications liberalization in the 1970s–1980s, financial deregulation in the 1980s–1990s, and recent deregulatory activity. We decompose the index by sector, type of deregulation, and policy stage. We validate the […]

  • AI and Coder Employment: Compiling the Evidence

    Leland D. Crane, Paul E. Soto

    We evaluate whether LLMs have had any discernible impact on the aggregate labor market so far. We focus on occupations that are computer programming-intensive, motivated by data showing that coding is one of the most LLM-exposed tasks. Linking O*NET to CPS we find that aggregate employment of coders has decelerated sharply since the introduction of ChatGPT. Using a novel control variable for industry-level shocks we show that the deceleration is not attributable to the exposure of coders to slowing industries, suggesting instead that coders experienced an occupation-specific shock around the introduction of ChatGPT. Coder employment has continued to grow in recent […]

  • The AI-GPR Index: Measuring Geopolitical Risk using Artificial Intelligence

    Matteo Iacoviello, Jonathan Tong

    We introduce an improved measure of geopolitical risk that builds on Caldara and Iacoviello (2022) and uses artificial intelligence to evaluate newspaper content. Our approach replaces keyword matching with semantic understanding: instead of searching for specific word combinations, we use one of the language models underlying ChatGPT (GPT-4o-mini) to read newspaper articles and assess their geopolitical risk intensity. The daily AI-GPR index scores about 5 million articles from the New York Times, Washington Post, and Chicago Tribune from 1960 through 2025. The approach reduces false positives from articles mentioning war or terrorism in non-geopolitical contexts while capturing relevant articles that lack […]

  • Artificial Intelligence Innovation by Financial Innovators: Evidence from US Patents

    Jean Xiao Timmerman

    This paper examines the evolution of artificial intelligence (AI) patent rates (i.e., the number of AI patents/number of firms of the same type) and concentration metrics (i.e., the Herfindahl-Hirschman Index (HHI) and Gini coefficient) among financial market participants from 2000 to 2020. It documents the historical trajectories of AI innovation for regulated banking entities and less-regulated firms, revealing that nonfinancial companies exhibit the highest baseline AI patent rate, while banks show the highest growth in AI patent rate over time. Banks have the highest HHI, and nonfinancial companies have the highest Gini coefficient, suggesting that a small number of banks dominate […]

  • The State of AI Competition in Advanced Economies

    Alex Haag

    Global competition in artificial intelligence (AI) has intensified in recent years. Some assessments emphasize US exceptionalism, while others argue that China is eroding US dominance. By contrast, the progress of other advanced foreign economies (AFEs) receives far less attention. Existing cross-country comparisons rely largely on composite indices that, while useful as benchmarks, are subject to weighting and aggregation biases that may obscure important dimensions of AI capacity. A clearer understanding of cross-country AI capabilities can help better contextualize global AI competition. This note brings together international comparisons on key metrics to assess countries’ relative preparedness and performance in AI. Read the […]

  • Estimating Aggregate Data Center Investment with Project-level Data

    Eirik Eylands Brandsaas, Daniel Garcia, Robert Kurtzman, Joseph Nichols, Adelia Zytek

    Data center investment in the U.S. has increased rapidly in the post-pandemic era, and plans for future investment have surged further. Forecasting investment at such a turning point is an important but potentially fraught exercise, especially given lags in aggregate data availability. We develop a straightforward method to forecast aggregate investment using project-level microdata and a small number of parameters: specifically, abandonment rates, time from plan-to-start, and time from start-to-completion. As a key validation of our approach, we generate estimates that match the recent history of aggregate data center investment in the NIPAs. We then use our method to generate nowcasts […]