Despite many decades of development, spatial prospectivity modelling is not yet widely used or accepted throughout the global mineral exploration industry. A common criticism of the method is that it is not practically useful because it has a bias to mature, well-known areas and generates excessively large areas of high-prospectivity. It is suggested that the reason for this is not primarily related to limitations in the prospectivity mapping algorithms but rather to issues relating to the use of input data sets. Specifically, it is common that the input data (such as geological interpretations) do not uniformly and objectively represent the search space of interest, omit critical targeting-relevant geoscientific elements (such as major, deep-seated ore-controlling structures) and have a large degree of unrecognised dependence.
It is considered that these problems are not in principle barriers to the eventual successful deployment of this technology. However, future approaches to spatial prospectivity modelling need to explicitly address these concerns. It is suggested that the most effective method may be a hybrid of subjective human geological interpretation and objective, machine-based analysis, that captures the best aspects of these alternative approaches; i.e., an intelligence amplification (IA) rather than an artificial intelligence (AI) approach. A roadmap is proposed for improving the effectiveness of spatial prospectivity modelling that has implications for the broader community interested in mineral exploration targeting.