AI is powering a new era of mineral discovery

How has a relatively unknown mining exploration company won the backing of billionaires Bill Gates and Jeff Bezos? It’s using artificial intelligence (AI) to hunt down critical minerals.

KoBold Metals, a California-based mineral exploration company that aims to drive the clean energy future by accelerating the discovery of critical metals, gathers data including satellite images and drilling results and applies AI to build “Google Maps” of the Earth’s crust in search for potential deposits of copper, lithium, cobalt and nickel. 

Earlier this year, the company made a major copper discovery using its AI technology. KoBold believes its Mingomba copper deposit in Zambia will be one of the world’s largest, high-grade copper mines.

Outside of Mingomba, the company is involved in exploring more than 60 projects, including the Solwezi copper project in Zambia, where it has partnered with Midnight Sun Mining Corp (TSX-V:MMA, OTCQB:MDNGF) to advance the project.

“By using AI at every step, [Kobold Metals] aim to cut exploration timelines from 10 years to three to four,” Midnight Sun Mining’s Adrian O’Brien told Proactive earlier this year.

Multimillion-dollar savings

Analysts, too, believe AI could transform resource discovery, which is known to be a high-fixed cost and manual labor-intensive business. 

“Compared to other industries such as business services and manufacturing which were early adopters of technological change, progressive productivity over time in mining has been very slow,” Jefferies analysts wrote in a note. 

“However, now with the industry's widespread adoption of AI technologies, mineral exploration and mine production will likely become more efficient, potentially leading to more commodity supply and lower operating costs.”

AI-powered data analysis and predictive technology use geological surveys, satellite imagery, and historical data to identify patterns and pinpoint target areas for exploration, improving efficiency and reducing exploration programs' cost and environmental impact. 

With exploration drilling productivity currently at its lowest due to rising costs, including higher labor and input costs, AI is becoming more appealing in early stage projects.

“AI technologies have the ability to collect and analyze large volumes of data at scale and speed in a standardized manner to improve the exploration success and lower operating costs,” Jefferies wrote. 

Capital expenditures on projects have also been depressed, as miners continue to operate in an age of austerity, the analysts noted.

“If the capital intensity and lead-time to deliver projects can go down as a result, we believe some projects that have been put on hold could become economically viable, leading to an increased supply of mined commodities.” 

Mining firms are forecast to spend $218 million on AI tools worldwide by 2024. This is up from $76 million in 2019, representing a compound annual growth rate (CAGR) of 23.4%. 

But this investment is expected to pay off in spades, with McKinsey estimating that by 2035, using AI will save miners between $290 billion and $390 billion annually. 

Leveraging AI for exploration wins

Driven by the promise of these benefits, majors and juniors alike are already deploying AI technologies in the field.

Last month, Barrick Gold (TSX:ABX, NYSE:GOLD) tapped Fleet Space Technologies to deploy its mineral exploration solution ExoSphere to advance exploration at its Reko Diq copper project in Pakistan. 

Fleet Space’s technology is being used to generate 3D subsurface maps of groundwater systems and copper porphyry complexes at multiple areas of interest spanning over 1,150 square kilometers of the project. 

Canadian firm Lancaster Resources (CSE:LCR, OTCQB:LANRF) has entered into an agreement with KorrAI to integrate hyper-spectral imaging and AI modelling technology into its exploration processes.

Recently announced exploration programs at its Catley Lake and Centennial East properties in Saskatchewan's Athabasca Basin will leverage KorrAI’s offering, including an initial detailed surface outcrop mapping, using state-of-the-art hyperspectral data to identify vegetation stress indicative of subsurface uranium, integrating bedrock geology, geological structures, and geophysics for field targeting. 

Over in British Columbia, Kodiak Copper (TSX-V:KDK, OTCQB:KDKCF) has leveraged VRIFY’s AI mineral targeting software to enhance its exploration efforts at the MPD copper-gold project. 

VRIFY’s AI targeting system builds predictive models using exploration data such as drill holes, rock geochemistry, and mineral occurrences. It pointed to drill targets already identified by the company in addition to new prospective areas. 

“The VRIFY AI software has proven a remarkable tool to support our ongoing exploration work, and its analytical power has contributed to defining the most promising targets for this year's exploration program,” Kodiak Copper CEO Claudia Tornquist told investors. 

Promising results

Meanwhile, ATHA Energy (CSE:SASK, OTCQB:SASKF) is developing a machine learning project at its Gemini uranium project in Saskatchewan. It aims to detect previously unknown targets derived from data compilation of surficial mapping and a variety of geophysical survey types, such as magnetics and gravity. The results are expected to guide an upcoming drill program. 

“The process allows us to efficiently, and with the highest probability, derisk and identify high-priority targets across our dominant land packages in the best uranium jurisdictions in Canada,” ATHA CEO Troy Boisjoli explained.

ATHA’s machine learning exploration work has been recognized as industry-leading, taking home the first place prize, Best Overall Award, at the Future Explorer’s competition, funded by Dundee Precious Metals (TSX:DPM) in June. 

The application of AI in mining is still in its early stages, but the results are already promising. As the technology matures, it has the potential to revolutionize mineral exploration and production, helping the industry meet global demands more efficiently.