Artificial intelligence is poised to dramatically reshape mineral exploration — potentially reducing the amount of exploratory drilling required by a factor of five, according to Jef Caers, Professor of Earth and Planetary Sciences at Stanford University. Caers made the remarks during a webinar hosted as part of the World Mining Congress 2026, explaining that AI-driven decision systems can guide exploration far more efficiently than traditional grid-based drilling strategies.
Unlike conventional methods that drill according to predetermined patterns and often result in redundant holes, the AI approach treats exploration as a sequential decision-making problem under uncertainty. The system uses machine learning models — similar in concept to “intelligent agents” used in autonomous vehicles — to plan drilling campaigns that deliberately test and discard incorrect geological hypotheses, then focus only on reducing uncertainty where it matters most. This strategy sharply cuts unnecessary drilling while preserving the detail needed to make informed “go-no-go” decisions about resource potential.
Caers explained that in traditional exploration, companies typically build a single model of the subsurface and commit to drilling based on that one interpretation. But if early assumptions are wrong, much of the drilling effort becomes inefficient. The AI-based system dynamically adjusts drilling targets as new information arrives, reducing geological uncertainty much faster and with fewer boreholes. This method not only reduces time and costs but could also minimise the environmental footprint of exploration activities.
The World Mining Congress 2026, set for June in Lima, Peru, will bring together mining leaders, technologists, and researchers to discuss innovations such as AI-powered mineral targeting and sustainability strategies. Caers’s comments reflect a broader trend: AI is increasingly viewed as a transformative tool in sectors like mining by enhancing decision-making and reducing capital-intensive activities like drilling.
