Kubaba Labs Proposes Causal AI Framework to Discover Deeply Buried Critical Mineral Deposits

Highlights

• Alexandra Roslin of Kubaba Labs proposes Causal AI to model why mineralization occurs, moving beyond pattern recognition and Bayesian methods used by KoBold Metals.
• The framework addresses three key exploration challenges: biased drilling records, misleading geophysical signals, and poor model transferability across geological basins.
• The paper presents a theoretical blueprint combining structural causal models, fluid-flow physics, and statistical bias correction—but no new deposit has been discovered using it.
• REEx notes that if causal modeling reduces failed drilling, it could significantly improve exploration economics for critical minerals like copper.
• Independent benchmarking and prospective field trials measuring drilling costs and discovery rates are needed before commercial effectiveness can be assessed.