Artificial intelligence is poised to upend one of the mining industry’s most expensive and time-consuming activities: mineral exploration drilling. According to Jef Caers, Professor of Earth and Planetary Sciences at Stanford University, AI-driven exploration strategies could reduce drilling requirements by as much as a factor of five, unlocking major savings in both capital and time.
Caers delivered the message during a recent webinar in the World Mining Congress 2026 series, arguing that artificial intelligence does far more than improve predictions. Properly deployed, he said, it fundamentally changes the logic of how exploration decisions are made.
For decades, mineral exploration has followed a familiar pattern. Geologists build a single deterministic subsurface model based on limited data, then drill extensively often on a fixed grid, to estimate grades and tonnage. The method is capital-intensive and, critically, assumes that the initial geological interpretation is broadly correct. When that assumption fails, drilling efficiency collapses.
Caers contends that AI offers a way out of this trap by reframing exploration as a problem of sequential decision-making under uncertainty. Rather than committing early to one geological story and drilling heavily to confirm it, AI systems act as ‘intelligent agents’ that continuously plan, learn, and adapt.
To make the idea tangible, Caers pointed to autonomous vehicles now operating on the streets of San Francisco. Like self-driving cars that constantly sense their surroundings and adjust their actions in real time, intelligent agents in exploration make drilling decisions while simultaneously optimizing what information should be collected next. The objective is not to confirm a preconceived model, but to challenge it.
In practice, this means drilling to falsify human-generated geological hypotheses as quickly and efficiently as possible. If a hypothesis proves wrong, the system pivots early, avoiding large amounts of wasted drilling. Only once uncertainty has been sufficiently reduced does the focus shift to defining grades and tonnage with greater confidence.
“All critical mineral supply chain challenges can be seen as sequential planning under uncertainty problems, starting with exploration,” Caers said, underscoring the broader implications of the approach. As demand for critical minerals accelerates driven by energy transition technologies and geopolitical pressures, the ability to make faster, better-informed exploration decisions could become a strategic advantage.
The promise, Caers argues, is not merely fewer drill holes, but better outcomes. By dynamically adjusting drilling strategies as new data emerges, AI can help companies reach clear ‘go or no-go’ decisions earlier in a project’s life, reducing financial risk and improving capital discipline.
For an industry long defined by expensive bets on imperfect information, the rise of intelligent, decision-making AI could mark a turning point shifting exploration from brute-force drilling toward a more surgical, information-driven future.
