OpenStreetMap AI-Generated Buildings in OpenStreetMap
Frequency of Use and Differences from Non-AI-Generated Buildings
A new study published in the International Journal of Digital Earth investigates the role of AI-generated building data in OpenStreetMap (OSM). The research examines the frequency, distribution, and modifications of AI-generated buildings compared to human-contributed data. It finds that while AI-generated buildings are often identified through source tags and contribute significantly to OSM, they are unevenly distributed, with 75% found in five countries. AI-generated data is modified less frequently, suggesting a lack of continuous updates, unlike human-driven mapping efforts. Despite generally matching the quality expectations of data providers, the study highlights that AI models still miss many buildings, indicating the need for further refinement. The paper calls for improvements in AI-assisted mapping and the development of standardized practices to enhance data quality and community engagement.