Pragmatic GeoAI: Geographic Information as Externalized Practice

Abstract Current artificial intelligence (AI) approaches to handle geographic information (GI) reveal a fatal blindness for the practices of exactly those sciences whose methodological agendas are taken over with earth-shattering speed. At same time, there is an apparent inability remove human from loop, despite repeated efforts. Even though no question that deep learning has large potential, example, automating classification methods in remote sensing or geocoding text, current GeoAI frequently fail deal pragmatic basis spatial information, including various data generation, conceptualization and use according some purpose. We argue this failure direct consequence predominance structuralist ideas about information. Structuralism inherently blind purposes any representation, therefore fails account required A turn overcome problem.