Maps, Geographic Information, and Generative AI

Abstract

The advent of image generative AI has introduced the possibility of generating maps from simple textual prompts. This paper explores the potential uses and challenges of employing image generative AI in research, as well as its effects on map understanding. Through experimental map creation using this AI, it was found that while the technology can plausibly produce world maps, it struggles with accurately drawing medium- to large-scale maps, such as those of city-scale or with detailed features, as of December 2023. The resulting maps resembled maps in design, yet they lacked geographic information. On the other hand, just as it is possible to partially redraw maps maintaining a certain level of quality, creating fictitious or fake maps proves easier than generating factual ones. While image generative AI is expected to be applied to research on landscape evaluation, there are concerns that the generation of non-existent or fake maps may have a negative impact on academic research. The impact of this new technology on our geographical imagination and spatial cognition remains uncertain, necessitating ongoing discussion.