These AI Algorithms Drove the Recent Arkansas Lithium Discovery

Dr. Sakharov: With AI transforming industries across the board, the recent discovery of substantial lithium reserves in Arkansas illustrates just how impactful this technology has become. As AI reshapes our world, we’re also integrating it into our own projects in a slightly different way: automating chemical processes to extract higher yields of critical minerals. Just a 10% increase in efficiency can significantly enhance project profitability and address key technological risks.

These AI Algorithms Drove the Recent Arkansas Lithium Discovery

A hitch in the giddy-up of electric vehicle adoption is the sheer volume of lithium required, but the recent discovery in southeastern Arkansas might change that.

Driving the news: A group of researchers at the USGS used an AI algorithm called a random forest to estimate the in-place volume of lithium in the Smackover formation in Arkansas. They estimated there was somewhere between five and 19 million tons of lithium there.

Why it matters: If this lithium is commercially viable, it would provide nine times the projected demand for EVs in 2030.

Additionally, the random forests technology they used to make the discovery was an interesting example of how AI is being used in upstream energy.

The random forests algorithm combines the output of multiple decision trees to reach a single result. Researchers tuned their model using the tidy-models framework in R, testing XGBoost, K-nearest neighbors, and random forest algorithms. Random forest algorithms proved to have the highest accuracy and lowest bias, making them the choice for training the final model and predicting lithium concentrations.

While the estimates are exciting, researchers caution that these are in-place estimates only, and further studies are needed to determine technical recoverability.