A new vein: AI in the mining industry

Contributed by mining optimisation and automation subject matter expert Jason Nitz

A venture capital firm recently asked me to provide expert advice on an investment plan for mining technology. They’d chosen several areas where mining technology had recently improved a process or system and completed a market analysis on the leading companies within those areas. Their intent was to invest in a small to medium-sized company to inject capital to improve their product or services, and then sell the improved company and its product or service to a large corporation, OEM, or even miner.

At the meeting we rated and ranked the various technologies they’d short-listed and then reviewed target companies currently operating in those technology pillars. At the end of the exercise, they asked me a question that caused me to pause for thought:

“Are there any other technologies out there that we’ve missed?”

I had to admit there was nothing out there that I knew of that was going to be the next “big thing” in mining. If there was, whoever was working on it was keeping it very quiet — something not easily done in the mining industry. Granted, there are moves towards seabed and space mining, though they are a spin on existing mining methods but in new and harsher environments, and are many, many years away from becoming economically viable. There’s battery electric vehicle (BEV) technology and green hydrogen, though these were technologies flagged as being low maturity with high barriers to market entry. But one thought did come to mind.

What is AI, really?

Artificial Intelligence, or AI, is a technology that isn’t readily associated with mining. Tech companies such as Google, Facebook, Microsoft and Apple are what comes to mind first, as is OpenAI and its ChatGPT product.

There are many perceptions of what AI is, with most reminiscent of Hollywood movies such as Terminator and other sci-fi stories. However, the truth about what AI is and does is generally much simpler.

The term “artificial intelligence” is a misnomer. John McCarthy coined the phrase in 1956 to differentiate what he was working on from other studies at the time focused on “thinking machines.” It can be argued that there’s very little intelligence currently involved in AI today, as intelligence is made up of factors such as reasoning, logic, self-awareness and perception. “Assistance,” rather than “intelligence,” would be the better descriptive.

Generative AI, like ChatGPT, where answers are “generated” is only one aspect of AI. There are other branches such as computer vision, machine learning and the use of deep, neural networks that could provide benefit to mining processes and systems.

AI is all around us and has been for some time — you probably use versions of AI every day without knowing it.  Remember all those Teams and Zoom meetings during the pandemic (and even still today)? The ability for those programs to blur your camera background is a form of AI called computer vision. Even Netflix’s recommendation tool that suggests what you might like to view next uses AI to review your viewing history and make an informed suggestion. Using those examples, most AI today is very narrow — that is, it can only do what it was programmed to achieve. For example, the process behind scanning your face to unlock your phone can’t be used to navigate you to the nearest service station, even though both can be done from the same device. The AI technology used for the two are very different and are not interchangeable.

Artificial General Intelligence (AGI) is what the AI industry is working hard to achieve. AGI is the next level of AI, where the two tasks used in the above example could be completed by the one application. To achieve this, AI would need to exhibit human-like intelligence, with the ability to self-learn and self-teach. Some researchers say that by the end of this decade it’s possible this will be achieved given the exponential increase in computing power through smaller and faster integrated circuits. Others believe it’ll be the end of the century before that’s possible. But for most readers, it’s more the here and now we’re interested in.

Mining’s technological revolution

The case for AI in mining is vast — any activity that generates large amounts of data and/or requires humans to do repetitive tasks is ripe for AI. This is where the “assistance” mentioned earlier comes into play.

Mining at its core produces a lot of data, sometimes overwhelming amounts. Any tool that can help sift through that data and highlight the nuggets (pun intended) is a tool worth using.

Machine learning and neural networks are two such aspects of AI that can tackle this challenge with comparative ease.

Take analysing drill core samples for example. This task is usually done by graduate geologists under the supervision of a senior geologist and can be hot, back-breaking work. Rows and rows of drill core trays need analysing, classifying and cataloguing, often in less-than-conformable environments. Using a hand-held magnifier, these geologists spend hours bent over trays looking for signs of ore, rock structures and other tell-tale signs that can help define an ore body. Vision is the primary human sense the geologists are using to complete this task.

AI has become extremely good at vision in the past decade. Computer vision is already making its way into mining, for example in processing plants, where a human would normally keep an eye on things. Outside of mining, self-driving cars rely on computer vision to keep the car (and occupants) safe whilst obeying the road rules. Combine computer vision with machine learning, which is essentially a way of ‘training’ AI to learn, and you have a strong case for investing in AI to perform the same drill core analysis task much quicker than the geologists, allowing them to be freed up to perform more value-add tasks. Luleå University of Technology in Sweden is currently working on this in conjunction with a technology vendor on a project called ML4DrillCore, a fusion of machine learning and computer vision for drill core analysis.

AI across the value chain

AI is revolutionising every stage of the mining value chain, from extraction to processing and shipping.

At mines, AI can drive the adoption of autonomous vehicles and machinery. With autonomous equipment, workers can avoid high-risk or remote areas, improving safety and efficiency for mine sites while also reducing potential human error.

BHP (ASX: BHP) says its WA Iron Ore (WAIO) uses AI as a decision support system at its mines.

WAIO is an integrated system of four processing hubs and five mining hubs, connected by more than 1,000km of rail and port facilities.

With such a complex operation, potentially thousands of touchpoints are controlled through a remote operations centre.

With thousands of touchpoints come an exponential number of decisions to be made by humans. BHP team members make the final decision, but the decisions are aided by AI systems and their information processing capabilities.

Also in the WA Pilbara region, BHP has eight automated shiploaders at its Port Hedland export facility. BHP says the eight shiploaders are responsible for loading about 1500 bulk ore carriers annually, exporting approximately 280mt of iron ore to global customers in 2021.

BHP says automating its shiploader facilities has increased production by more than one million tonnes each year, through greater precision, reduced spillage, faster load times and equipment optimisation.

Additionally, in 2024, all mine trucks at BHP’s Spence operation in Chile were made fully autonomous, which BHP says unlocks safer and more efficient operations at site.

In his 2023 book The Coming Wave, Mustafa Suleyman speaks of technology coming in waves with history demonstrating there’s been several waves over the past 500 years, with Gutenberg’s printing press and Lenoir’s internal combustion engine starting waves that still resonate today. AI in mining may not be so much a wave, but the tide is certainly turning.

Stay tuned, as transformational digital expert Jason Nitz and the Australian Mining Review highlight AI developments across the entire mining value chain in future editions.

Jason Nitz has held a range of senior leadership positions in operations, technical services and technology, having worked both in Australia and overseas for several global mining companies over the past 21 years. He is currently focused on working with mining companies to optimise their operations through the introduction of transformational digital solutions, including the incorporation and use of AI, along with educating the wider mining industry on the adoption of AI and its benefits.