On The Rocks Exploration AI 101: Transforming Uncertainty in Mineral Discovery
Jun 17, 2025
Join John Mern, CEO of TerraAI and a Stanford PhD, and Luke Ren, Chief of Staff with a savvy investment background, as they unravel the transformative powers of AI in mineral exploration. They tackle the prevalent uncertainties in the industry, reveal how AI enhances traditional methods, and discuss the role of investment dynamics influenced by data-driven strategies. The duo also highlights innovative approaches to risk management and how emerging technologies can attract new investments, making mining more efficient and viable.
AI Snips
Chapters
Transcript
Episode notes
Subsurface Uncertainty Is Orders Harder
- Modeling subsurface uncertainty is far harder than typical engineering problems due to scale and dimensionality.
- Mines require millions of parameters versus a few numbers for systems like aircraft, making traditional uncertainty models insufficient.
Fuse All Data To Reduce Incompleteness
- Use all available data types together instead of single-mode analyses to reduce incompleteness in uncertainty models.
- Combine geophysics, geochemistry, geology and expert hypotheses to produce honest, constrained uncertainty estimates.
One Guess Plus A Cone Misleads
- Simple confidence cones around a single best-guess model dramatically understate subsurface uncertainty.
- Multiple, vastly different explanations can fit the same signal, so honest distributional uncertainty matters.