Geographia Technica, Vol 21(2), Special Issue: Artificial Intelligence Applications in Geography, 2026, pp. 230-241
ARTIFICIAL INTELLIGENCE IN GEOGRAPHICAL RESEARCH: FROM SPATIAL OBSERVATION TO INFORMED ACTION - A SYNTHESIS OF THE AIAG STUDIES AND THE GEOGRAPHICAL FOUNDATIONS OF GEOAI
Zsolt MAGYARI-SÁSKA 
ABSTRACT: This paper synthesises the twelve contributions included in the AIAG special issue and considers their findings within the broader development of geospatial artificial intelligence. The studies show how AI and related computational methods can support the identification of geographical phenomena, the estimation of environmental conditions in data-sparse areas, the analysis of spatiotemporal processes, the construction of land-use and hazard scenarios, and the integration of data-driven models with physical knowledge. Rather than treating algorithms as the principal outcome, the synthesis focuses on the geographical evidence they produce, including flood boundaries, pollution estimates, rainfall forecasts, biomass inventories, burned-area maps, land-use projections and susceptibility assessments. The discussion then shifts from what AI performs in the individual studies to what geography contributes to AI. It examines how geographical research defines meaningful entities, represents scale, connectivity and temporal structure, organises heterogeneous evidence, keeps environmental and social processes visible, and connects model outputs to places, institutions and decisions. These contributions are essential because spatial coordinates alone do not make a model geographically meaningful. The paper also considers the conditions required for reliable GeoAI, particularly spatially appropriate validation, transferability between regions, interpretability and the spatial communication of uncertainty. It argues that AI is most useful when it complements GIS, remote sensing, field observations, statistical analysis, physical modelling and expert knowledge. Its contribution lies not in replacing geographical judgement, but in extending the capacity to observe, estimate and anticipate spatial processes while keeping model outputs connected to the geographical contexts and decisions they represent.
Keywords: GeoAI; geographical knowledge; spatial dependence; geographical modelling; decision support; spatial validation; explainable AI
Full article here