Using geospatial artificial intelligence to advance the Sustainable Development Goals

• GeoAI task types were summarized based on AI task types. • GeoAI tasks in SDGs were matched. • A practical framework for GeoAI-enabled SDGs achievement was proposed. • Approaches for integrating machine learning with geography are elucidated. • Limitations and future strategies for GeoAI-enabled SDGs achievement were proposed. The Sustainable Development Goals (SDGs) have become the focus of sustainable science research since it was proposed by the United Nations in 2015. Despite SDGs have made significant progress, it remains challenging to fully realize them by 2030. Compared with traditional geographical methods, geospatial artificial intelligence (GeoAI) method has shown significant technical advantages. However, research on the application of GeoAI in the SDG implementation remains limited. We argue that GeoAI can greatly promote SDG implementation under the premise of standardized application. Specifically, we classify GeoAI task types into GeoAI classification (GeoAI C) and GeoAI regression (GeoAI-R) based on AI task types, and systematically evaluates SDG-related GeoAI research objects and their research tasks. We further outline the practical path of using GeoAI to advance SDGs from the aspects of target determination, data preparation and model construction. We also discuss the limitations of the method system and future research strategies. This article provides theoretical guidance for integrating GeoAI into efforts to achieve the SDGs.