Beyond “Geo” HCI: Exploring Cultural Dimensions of Disparity in OpenStreetMap Road Safety Metadata

Peer production systems like OpenStreetMap exhibit well-known information gaps along socioeconomic and population density lines, causing issues for end users when AI tools, such as autonomous vehicles, rely on this incomplete data. Prior work has shown these trends, but how they influence important semantic metadata remains unclear. In this study, we focus on three OpenStreetMap metadata tags that are essential for road safety. Contrary to the expected socioeconomic and population density trends, our findings reveal that cultural factors play a significant role in influencing tag production. Moreover, we find that automated contributions can negatively impact human tag production in OpenStreetMap, despite the potential of these automated imports to address data gaps. Our results add nuance to the trade-offs of automated imports and shed light on how the public and practitioners can more effectively improve metadata coverage.