Geospatial science refers to the interdisciplinary field involving acquisition, analysis, modelling, interpretation, visualization and service of geospatial data related to Earth’s surface and human activities. It has, since the 1960s, undergone an analogue-to-digital transformation (or digitalization) and is currently experiencing a digital-to-intelligent transformation (or intelligentalization). The Spring Equinox of such a transformation might be the launch of GPT-4 in 2023, which is regarded as the spark of AGI (artificial general intelligence). However, current LLMs like GPT-4 remain as early (yet still incomplete) versions of AGI, struggling in addressing geospatial domain-specific problems. Therefore, the hybrid integration of AI with geospatial-NI (natural intelligence in geospatial science) to form geospatial-NI-augmented AI might be the only feasible solution at this development stage. In such an AI system, the NI in the form of geospatial knowledge plays a vital role, while the hybrid computing model in the form of KDAS (knowledge-guided, data-driven, algorithm/model-based, and service-supported) serves as the fundamental paradigm. A number of challenges will be faced, and research issues should be tackled during the intelligentalization process. The further development of AI-empowered geospatial science will drive a shift toward “wisdomalization”, elevating the field from knowledge-guided to wisdom-guided.
Li et al. (Wed,) studied this question.