Experiencing Uncertainty: The Introduction of Liminality and Latency in Speculative Mapping of Lake Karla

GeoAI technologies have expanded the analytic power of cartography by processing vast geospatial datasetsto detect patterns, generate multiscale maps, and unify siloed information. Despite their innovation, suchsystems often reinforce static spatial conventions, treating maps as neutral outputs and uncertainty, whetherwithin datasets or landscapes, as an issue to be resolved. This paper calls for a critical reorientation ofGeoAI’s epistemological foundation, proposing that uncertainty be understood not as a limitation but as thefoundation for grounding new protocols of spatial representation. Building on the two key concepts of latencyand liminality, we develop a framework for understanding uncertainty as both an epistemic gap and anontological threshold. These concepts are operationalized through three architectural drawing strategies:concealing, completing, and abstracting. Lake Karla, a climate-sensitive and politically contested landscapein central Greece, serves as a case study. Using a custom-designed radial diagram and hybrid visualtechniques, we integrate temporal and spatial data with AI-assisted media analysis to visualize the lake’sshifting identity. Our findings suggest that Lake Karla’s transformations between natural, artificial, andambiguous states cannot be captured by fixed cartographic classifications. Instead, uncertainty emerges asa core spatial feature.