Generative artificial intelligence systems based on transformers, including both text-generators like GPT-4 and image generators like DALL-E 3, have recently entered the popular consciousness. These tools, while impressive, are liable to reproduce, exacerbate, and reinforce extant human social biases, such as gender and racial biases. In this paper, we systematically review the extent to which DALL-E Mini suffers from this problem. In line with the Model Card published alongside DALL-E Mini by its creators, we find that the images it produces (...) tend to represent dozens of different occupations as populated either solely by men (e.g., pilot, builder, plumber) or solely by women (e.g., hairdresser, receptionist, dietitian). In addition, the images DALL-E Mini produces tend to represent most occupations as populated primarily or solely by White people (e.g., farmer, painter, prison officer, software engineer) and very few by non-White people (e.g., pastor, rapper). These findings suggest that exciting new AI technologies should be critically scrutinized and perhaps regulated before they are unleashed on society. (shrink)

The COVID-19 pandemic has upended in-person public education across the United States, a critical infrastructure of care that parents—especially mothers—depend on to work. To understand the nature and magnitude of school closures across states, we collected detailed primary data—the Elementary School Operating Status database —to measure the percentage of school districts offering in-person, remote, and hybrid instruction models for elementary schools by state in September 2020. We link these data to the Current Population Survey to evaluate the association between school (...) reopening and parents’ labor force participation rates, comparing 2020 labor force participation rates to those observed prepandemic in 2019. We find that, across states, the maternal labor force participation rate fell to a greater extent than that of fathers. In 2019, mothers’ rate of labor force participation was about 18 percentage points lower than fathers’. By 2020, this gap grew by 5 percentage points in states where schools offered primarily remote instruction. We show that schools are a vital source of care for young children, and that without in-person instruction, mothers have been sidelined from the labor force. The longer these conditions remain in place, the more difficult it may be for mothers to fully recover from prolonged spells of nonemployment, resulting in reduced occupational opportunities and lifetime earnings. (shrink)

In this article, we investigate how state-to-state differences in U.S. childcare costs and gender norms are associated with maternal employment. Although an abundance of research has examined factors that influence mothers’ employment, few studies explore the interrelationship between maternal employment and culture, policy, and individual resources across U.S. states. Using a representative sample of women in the 2017 American Community Survey along with state-level measures of childcare costs and gender norms, we examine the relationship between these state conditions and mothers’ (...) probability of employment. We pay careful attention to differences in mothers’ level of education. Our results suggest that expensive childcare is associated with lower maternal employment, particularly for those with less education. For the college educated, expensive childcare is negatively associated with maternal employment in states with traditional gender norms that uphold mothers as primary caregivers. Among mothers with lower levels of education, gender norms have a limited association with employment. These findings suggest that highly educated mothers mobilize resources to remain in the labor force when paid work is supported by local gender norms. For less-educated mothers, expensive childcare predicts lower employment regardless of gender norms, indicating that structural constraints outweigh normative expectations among those with fewer resources. (shrink)