Application Of Artificial Intelligence In Mineral Exploration – Case Study Cajamarca, Peru P.Geo., Eng., Msc., Santiago Mayor1, P.Geo., Eng., Msc., Hector Canales2, P.Geo.,P.Eng., Msc., Andre Gauthier3 Geomatic World Inc, Montréal, Canada, (*Presenting author: santiago_mayor@geomaticworld.com) Abstract Mineral resources are essential for economic and social development, yet the discovery rate of new deposits has decreased, particularly for large deposits that contain most of the world’s metal endowment. As global demand continues to grow, exploration must increasingly focus on underexplored areas, where geological complexity, limited access, and environmental constraints increase risk and uncertainty. This study presents an artificial-intelligence-driven predictive analysis workflow designed to reduce exploration risk and improve decision-making in the Cajamarca region of northern Peru. Advanced geospatial datasets and geographic information system (GIS) workflows were combined with Weights of Evidence, Fuzzy Logic, and Machine Learning techniques to integrate geological, structural, geochemical, geophysical, topographic, and remote-sensing information into a Mineral Prospectivity Map (MPM). The methodology included extensive data preprocessing, cleaning erroneous or missing values, applying logarithmic and exponential transformations where justified, and normalizing heterogeneous variables to a common scale. Discrete variables such as lithology and faults were converted to continuous representations to enable consistent modeling. Fuzzy Logic allowed the inclusion of expert knowledge by manually adjusting the relative weights of key variables and deposit-related evidential layers, while supervised Machine Learning algorithms were calibrated and evaluated using F1-Score metrics to balance false positives and false negatives. Results show that areas where multiple favorable geological, geochemical, geophysical, and remote-sensing criteria overlap systematically achieve the highest prospectivity scores and coincide with the most known deposits in the region. The workflow also highlighted new priority targets that were validated in collaboration with IAMGOLD Perú through field sampling and spectral analysis, although detailed results remain confidential. The AI-based approach proved particularly effective in overcoming challenges associated with dense vegetation and rugged terrain by integrating passive sensors (ASTER, WorldView) with radiometric and geochemical data. The study demonstrates that AI-driven prospectivity modeling is a robust, scalable, and transferable tool for responsible mineral exploration. By reducing uncertainty, optimizing target definition, and improving resource allocation, the proposed workflow can potentially contribute to delivering the minerals society urgently needs in a more efficient and sustainable way. 198
RkJQdWJsaXNoZXIy MTM0Mzk2