SMART TARGET DETECTION IN MINERAL EXPLORATION: HARNESSING AI UNDER UNCERTAINTY *B. Sadeghi1 1 Earth and Sustainability Science Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, NSW 2052, Australia, (*Presenting author: z5218858@zmail.unsw.edu.au) ABSTRACT Mineral exploration is inherently uncertain, with high investment costs and limited chances of discovery. Traditional decision-making approaches often focus either on geological potential or financial feasibility, but rarely integrate both in a rigorous, quantitative framework. To address this gap, we introduce an AI- and data-driven Return–Risk decision-making method that balances exploration success potential against geological, technical, and financial uncertainties. The proposed framework integrates heterogeneous datasets—including geochemical, geophysical, geological, and remote sensing information—with probabilistic models and machine learning techniques, with this study emphasizing geochemical data. The approach combines frequentist and Bayesian principles to generate an optimized strategy for identifying targets that maximize element concentrations (defined as “Return”) while minimizing uncertainty, thereby improving confidence that geochemical signatures genuinely reflect mineralization influence. These measures are synthesized into a Return–Risk matrix, which enables prioritization of prospects not only by their expected reward but also by their associated uncertainty profile. Case studies demonstrate that this approach enhances decision transparency by explicitly showing trade-offs between high-return but high-risk prospects and lower-return but more reliable opportunities. By embedding explainable AI into the workflow, geoscientists and decision-makers gain interpretable insights into why a prospect is ranked as high or low priority. This is particularly critical in frontier terrains, where investment decisions must be justified despite limited data. The Return–Risk method provides a bridge between geoscientific knowledge, advanced statistics, and economic reasoning. It allows stakeholders to optimize exploration portfolios, allocate budgets more effectively, and reduce decision bias. The framework is designed to be flexible and scalable, with applications ranging from greenfield regional targeting to advanced-stage prospect ranking. This contribution highlights how coupling AI with uncertainty quantification can transform mineral exploration into a more data-driven, transparent, and economically grounded process. Ultimately, the Return–Risk decision-making method empowers exploration companies to move beyond intuition-driven targeting toward systematic approaches that maximize discovery success while managing risk exposure. KEYWORDS Target detection, Decision-making, Uncertainty, Mineral Exploration
RkJQdWJsaXNoZXIy MTM0Mzk2