We study whether, and how, societal views about which problems are most important or worth solving are reflected in the direction of technological progress. Compared to Republican inventors, Democrats are more likely to patent technologies addressing climate-change mitigation or women’s reproductive health and less likely to patent weapons. These gaps are not explained by differences in inventive ability, persist within organizations, have strengthened over the past two decades, and also appear in subsequent innovation. Together, societal views and their polarization appear to be important drivers of technological change, operating in part through inventors’ technology choices, with implications for innovation policy.

International organisations devote substantial resources to reaching the poorest segments of the population. Yet we know little about how these resources are distributed and why. In this paper, we provide the first systematic evidence on where development programs go within countries. We assemble a new dataset linking roughly 5,000 geocoded World Bank projects across 130 countries since 1995 to district-level nighttime lights and complete records of the personnel assigned to each project. We find that aid flows disproportionately to richer areas: 51% of World Bank projects exhibit positive “site selection,” landing in districts significantly richer than their neighbours in the same province. In contrast, only about one-third of World Bank projects are targeted to poorer districts. The pattern is pervasive across sectors and is not explained by population-based targeting. It does not reflect a gain in performance: projects in richer areas are no more successful according to the World Bank’s own independent ratings. Supply-side factors related to World Bank personnel help explain where projects are located. World Bank project managers, who have substantial discretion over choosing project locations, account for a sizeable share of the total variation in site selection, more than year, sector, and country fixed effects combined. We unpack the role of career incentives, managerial skills, and learning in driving site selection. These findings suggest that where aid goes within countries is shaped not only by the needs of the recipient country but also by the organizational structure of the agencies that deliver it.

AI-led interviews are a key tool for understanding people’s views and reasoning at scale. This paper provides guidance on when and how to use them in online surveys, highlighting their potential benefits and costs.