Logo image
Back
Towards the Uncertainty-aware Geospatial Artificial Intelligence
Conference proceeding

Towards the Uncertainty-aware Geospatial Artificial Intelligence

Xiayin Lou and Peng Luo
Proceedings of the 8th ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery, pp.76-80
ACM Conferences
GeoAI '25: 8th ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery
11/03/2025
DOI: 10.1145/3764912.3770821
url
https://doi.org/10.1145/3764912.3770821View
Published (Version of record) Open Access

Abstract

Addressing geospatial problems with Geospatial artificial intelligence (GeoAI) is a growing trend. Combining spatial theory and cutting-edge AI models, GeoAI models have achieved far greater performance (e.g., accuracy) than traditional models. However, inherent uncertainty exists in GeoAI-based studies—from data collection to model construction and outputs—yet it has rarely been systematically considered or discussed. The existence of uncertainty can pose a threat to trust in the GeoAI models and even cause geographical bias. In this work, we investigate routes to develop an uncertainty-aware framework for GeoAI. We first categorize the major methods for quantifying uncertainty and their pros and cons. Then, the challenges for limiting the employment of uncertainty quantification methods in a geospatial context are described in detail. In the face of these challenges, we discuss the general characteristics of uncertainty-aware GeoAI that it should be equipped with. The introduction of geographical uncertainty could lead to more reliable and explainable GeoAI models while mitigating geographical bias.
 Applied computing -- Physical sciences and engineering -- Earth and atmospheric sciences Computing methodologies -- Artificial intelligence Computing methodologies -- Modeling and simulation -- Model development and analysis -- Uncertainty quantification

Details

Metrics

1 Record Views
1 Times Cited - Web of Science
Logo image