Cookie Preferences

AI to Unlock Exploration Frontiers

Artificial intelligence (AI) could transform mining exploration this decade, enabling companies to analyse larger datasets and develop new geological insights.

That’s the view of Ben Jupp and Stephen Johnson, respectively principal and senior geologists at SRK Consulting, an international resources consultancy.

They believe machine learning in mineral exploration, while still an emerging technology, will become more widespread as miners and explorers use algorithms to test and refine geological concepts currently applied to exploration.

One key area in particular is the application of AI in prospectivity analysis, an area in which SRK has recently seen success.

“I’m an absolute convert to using AI in mineral exploration,” Johnson said. “I’ve seen the benefits of using machine learning first-hand through a number of recent projects. The results have been incredibly powerful.”

For Jupp, the key is integrating AI with existing knowledge-driven exploration approaches.

“Essentially, AI allows us to integrate deep technical knowledge about a particular terrain and mineral system into a set of critical inputs to provide a machine learning algorithm,” Jupp said.

Mineral exploration methods are evolving with the advent of new technologies.
Mineral exploration methods are evolving with the advent of new technologies.

An advantage of using AI in prospectivity analysis is the rapid integration and analysis of large datasets by machine learning.

“Using training data such as known mines, machine learning can identify patterns and relationships in the data that geologists might overlook with traditional approaches,” Jupp said.

“Often, exploration companies will pay a highly knowledgeable expert to assist with their exploration targeting and narrow down their exploration efforts. With AI, we can use that knowledge, insight and human expertise to map out critical targeting elements and feed this into the machine learning to analyse and make predictions.”

AI is especially powerful for small and mid-sized mining companies that need to analyse large volumes of data cost-effectively.

“For junior explorers with a package of tenements and limited capital, the ability to narrow exploration targeting is crucial,” Johnson said. “AI can help do that and potentially find new value in tenements that have been overlooked or are a lower priority.”

AI also removes human biases and assumptions in exploration targeting.

“A consistent theme in exploration projects is people being drawn to certain areas based on existing data and preconceived ideas,” Johnson said. “With AI targeting, you’re feeding the available data and geological layers into a model to develop a holistic view of the prospectivity of a tenement package.

“There’s no bias or subjectivity about where you should be exploring beyond what the data is suggesting.”

AI projects underway

SRK’s interest in AI emerged around 2019. Jupp and Johnson were part of a team of Australian and Canadian SRK Consulting experts who competed in OZ Minerals’ Explorer Challenge, a prominent competition that attracted more than 1000 participants from 62 countries.

The SRK Consulting team won the Fusion Prize after reinterpreting and adding value to existing datasets by applying data-driven machine learning to guide a set of knowledge-driven, mineral-system-informed fuzzy inference solutions. The result was three highly ranked iron-oxide copper gold (IOCG) targets and seven secondary targets.

Data inputs from left to right, used to inform a machine learning generated prospectivity map. Image: SRK Consulting
Data inputs from left to right, used to inform a machine learning generated prospectivity map.

In 2021, SRK began applying these and other machine-learning techniques to help companies reduce their exploration targeting at brownfield and greenfield sites.

SRK has a global partnership with DeepIQ, a leading US developer of generative AI in the oil and gas, utilities and mining sectors. SRK experts have integrated DeepIQ algorithms into their mineral prospectivity analysis at some projects.

SRK has also recently applied machine learning to exploration projects in several regional project areas, including Australia, South America and Europe, with good results.

“There’s a lot of work underway at SRK globally to test and apply AI methodology for mineral exploration,” Jupp said.

Knowledge-driven targeting

SRK’s work with machine learning builds on its work in prospectivity analysis using more traditional knowledge-driven approaches for developing prospectivity maps. Methods such as fuzzy logic in mineral prospectivity analysis aim to quantify intricate relationships between geological attributes to define mineralisation potential.

“The main aim of the prospectivity analysis process is to assist our clients to narrow the focus area for exploration prior to field-based exploration, ideally as specific as the prospect area and even drill target areas,” Jupp said. “One of the key benefits of this method is you’re not reliant on the availability of training data when compared to machine learning methods.”

SRK recently applied fuzzy logic targeting on a project for Astute Metals, an ASX-listed resource company. Astute holds an 80 per cent interest in the Georgina Basin IOCG project in the highly prospective east Tennant province of the Northern Territory.

A machine learning predicted prospectivity map (left) and a structural feature input layer (right).
A machine learning predicted prospectivity map (left) and a structural feature input layer (right).

“We were successful in identifying several promising target areas in the undercover extents of east Tennant Creek,” Jupp said. “Recent drilling by the client at one of these targets intercepted strong indications for IOCG-style mineralisation that will be followed up with additional drilling to be completed later this year.”

In the Middle East, Johnson has used fuzzy logic processes for exploration targeting.

“Through our fuzzy logic approach, we developed a prospectivity model that we followed up with systematic fieldwork to validate,” Johnson said. “The outcome was fantastic because the fuzzy logic results were able to be tested and refined during follow-up phases of field work, with some really encouraging results.”

Bright future for AI

Jupp and Johnson believe industry hesitance towards using AI in exploration will fade as more results are proven.

“Some explorers have resisted the technology due to concerns that AI could overlook critical steps in the discovery process, but that will change as the industry sees meaningful exploration results from AI,” Johnson said.

“We’ve been getting a lot more enquiries on AI in exploration as interest in this area builds.”

Longer term, Johnson likes AI’s potential to challenge exploration “dogmas”.

“The mining industry has traditionally been influenced by certain dogmas about how and where exploration should be undertaken,” he said.

“Then occasionally someone makes a significant discovery that challenges this dogma and there is a rush to embrace new thinking. That will be true of AI this decade as it contributes to significant new mineral discoveries, and some early adopters are big winners from the AI revolution.”

This feature appeared in the August 2024 issue of Australian Mining.