By Matthew Blenkarn
To unlock the true value of artificial intelligence and machine learning, technology teams need to understand the particular benefits of each technology.
It’s not always easy to tell the difference between Artificial Intelligence (AI) and machine learning in mining. The famous computer scientist John McCarthy neatly described AI as ‘the science and engineering of making intelligent machines.’ Today, that definition neatly fits a number of the tools that metals and mining companies use to carry out crucial tasks. From systems that increase mill throughput to predictive maintenance platforms, there are a host of solutions across the value chain that leverage AI and machine learning to repair, preserve and optimise traditional mining equipment.
In fact, these two fields are closely interconnected within metals and mining. AI can broadly be understood as technology that replicates and expands upon human intelligence, encompassing everything from visual perception to pattern recognition to data analysis. Machine learning is just one of these subsets. As the name suggests, it enables technology to learn from data in an unassisted, automatic way, without the limitations or rules other areas of artificial intelligence need to achieve outcomes.
But while these fields are interrelated, they can be used for very different purposes across the value chain. That’s why it’s important to understand the differences between AI and machine learning in metals and mining order to unlock the true value of each.
What is artificial intelligence?
As previously suggested, artificial intelligence is less an individual technology and more of a broad category that encompasses many different functions and applications. This can lead to misconceptions that affect many industries, and mining is no exception. On a recent episode of Axora’s Innovation Digest audio series, Conundrum CEO Konstantin Kiselev noted that while many professionals conceive of AI in a broad, generalised way, its actual use cases are far more specific.
“Generalised artificial intelligence… is more from sci-fi movies,” he added. “Nobody knows how to implement general artificial intelligence, and there are many discussions about how it could be implemented.”
“AI systems can identify risks and analyse data patterns to find cost saving opportunities that human beings might miss”
Furthermore, the use of AI in metals and mining remains highly specialised. In order to deploy AI-based technologies, the industry needs robust and scalable infrastructure, the likes of which are not currently available at many sites. So even if technology departments had a way to implement generalised artificial intelligence, their operations would probably not be digitally mature enough to support it.
With that in mind, the use of specialised artificial intelligence has risen substantially within the industry over the past decade. Kiselev points to autonomous vehicles as a key example. Where human operators used to handle a narrow set of specialised tasks, AI-enabled machinery can now achieve the visual perception and pattern recognition needed to transport materials from one area to another. The same technology applies to areas like supply chain management, where AI systems can identify risks and analyse data patterns to find cost saving opportunities that human beings might miss.
What is machine learning?
While artificial intelligence can analyse data and recognise patterns faster than humans, machine learning compiles those insights and uses them to automatically adjust the way a process functions. As a result, technology that leverages machine learning is often extremely valuable for identifying and eliminating operational inefficiencies. For example, solutions that monitor material flow in mills have proven to be valuable in both increasing throughput and reducing energy consumption.
Exploration and mineral discovery have also benefitted substantially from the predictive power of machine learning in recent years. With enough data, a solution can effectively highlight points of interest such as metal ore bodies and alteration haloes with greater accuracy. When it comes time to blast rock, the same technology can use variables like explosive type and surface hardness to both minimise explosive use and increase the lifespan of drilling equipment.
…Industry leaders are using AI and machine learning to boost bottom-line success, while also doubling down on worker safety, mineral exploration, extraction, and transportation, and reducing ecological impact
Machine learning’s predictive capabilities are especially pronounced when it comes to maintenance applications, which enhance both efficiency and overall health and safety. By processing and reviewing data from sources like incident reports and root cause analysis, the technology can predict equipment failure with a very high degree of accuracy. For example, a solution like this could be used to pre-empt and prevent costly delays caused by a broken-down truck blocking a main haul road.
Environmental monitoring is another area in which machine learning is having a positive impact within the industry. The technology can be used to analyse data gathered by sensors and tracking systems to gain important insights relating to factors like ground water, subterranean ventilation, and temperature. The result of this is an enhanced understanding of potentially dangerous conditions like erosion, or wildlife habitats that must be protected.
Unlocking the benefits
While AI and machine learning share underlying functions, they offer distinct capabilities and benefits. Like any technology, neither offers a blanket solution to a specific problem. Without being applied to a specific challenge with a particular purpose, artificial intelligence and machine learning will provide little benefit.
So where does each technology go from here? While it may still be early days for AI in metals and mining, Kiselev predicts that it will gain a sizable foothold in processing plants. These environments involve so many variables and parameters that human operators can only maintain limited control over operations. With AI systems in place, plant staff can more accurately monitor critical properties like ore hardness and weight.
Organisations that have led the way and are already experiencing the benefits of these technologies include the likes of Rio Tinto, BHP, Barrick, and Freeport McMoran. These industry leaders are using AI and machine learning to boost bottom-line success, while also doubling down on worker safety, mineral exploration, extraction, and transportation, and reducing ecological impact. While the true benefits of AI and machine learning have yet to be achieved, their promise remains vast.
Matthew Blenkarn is a content producer at AXORA