Breaking new ground: Why the generative AI revolution matters for mining

Just as the internet and smartphones did before it, generative AI is poised to revolutionize the way we live and work. But what are the implications for mining companies?

There’s no doubt senior leaders across the mining industry are thinking hard about how and where to apply this technology. And many employees are already experimenting with ChatGPT in their daily work.

Indeed, this is one of the revolutionary aspects of generative AI—its accessibility and ease of use mean anyone can get started straight away.

But business leaders recognize the risks in letting these experiments run away uncontrolled.

There are real and genuine issues—around validating output accuracy and data security in particular—that need to be addressed.

Those leaders and their teams ?need to be clear what generative AI can and can’t do. It’s an incredibly powerful and exciting technology. But it’s not the answer to every business problem.

A carefully targeted roadmap focused on applicability to business processes, with use cases prioritized around business value and appropriate guardrails, is therefore essential.

A different kind of AI

So where do we see generative AI delivering that value in the mining industry?

What’s interesting is that there’s broad applicability across nearly all aspects of mining, from exploration to operations to supply chain to corporate functions and beyond.

That’s because, unlike traditional machine learning, generative AI isn’t only about deriving insight from data, but also improving processes and reinventing work.

Any language-based activities, or tasks that involve searching, summarizing or analyzing data, can potentially be augmented and accelerated with this technology.

This is all the more important given the industry’s workforce issues. As experienced personnel retire, companies are grappling with the challenges of upskilling the next generation of workers.

Ensuring safety

Worker safety is a key example. It goes without saying this is the number one priority in the industry. But, for new recruits, wading through mountains of safety documentation can be a steep learning curve and very time-consuming.

Even for experienced engineers, the volume of safety and other equipment information involved in carrying out specific repairs and maintenance can be onerous.

Generative AI’s advanced language abilities mean it can summarize all this documentation and describe the key points in an easily digestible format.

This can include everything from producing personalized training materials to providing real-time support on the ground through virtual assistants.

Accessing insights

There are countless other situations where generative AI can act as a friendly search engine, helping workers retrieve relevant insights from data faster, and in a form they can actually use quickly.

Take equipment failure. Right now, understanding the root causes can be very difficult to achieve because the relevant data is usually dispersed across numerous sources, documents and formats.?

But generative AI can process all these data sources—production logs, equipment readings, weather and environmental conditions, shift patterns, and more—to quickly suggest correlations and patterns that human analysis would take a long time to identify. The human can then go direct to the relevant data and validate the suggested correlations.

Accelerating processes

Then, there’s the impact in reporting. Today, mining employees create countless reports, slide decks and spreadsheets every week.

These take time to produce and are often filed away, never to be referred to again.

Generative AI can be used to not only accelerate the drafting of these documents, but also unearth new insights by comparing and correlating the wealth of data locked away in long-forgotten historical reports.

Expediting exploration

We also see many interesting use cases in exploration. That includes using generative AI to support the complex and document-heavy process of permitting and licensing.

And there’s potential to unlock remote sensing and legacy data to accelerate the search for prospective ground.

Consider that geoscientists commonly say they spend up to 70% of their time trying to identify, locate, access and compile data into a usable format, rather than analyzing it.?

Generative AI can cut this time, helping to identify the potential for mineralization in a given area more efficiently. That will ultimately reduce early-stage exploration expenditure while increasing the likelihood of discovery.

Understanding the limits

These are just a handful of the promising use cases across mining.?As we think about scaling them up, it’s important to be clear about some of the limitations of this technology.

Accuracy is a prime example. A generative AI model can—and will—invent information in order to produce a response to the prompt it’s been given.

That’s because, when you ask it a question, what you get back is not fact, but something that is statistically likely to be a suitable response.

This could still be incredibly useful in many ways, certainly in speeding up the process to get to the right answer. But it’s very different from having something that’s 100% guaranteed to be true.

There are also important business implications. You couldn’t, for instance, rely on generative AI to draft and then submit a regulatory filing on its own. It will always require human expertise to review, check, verify, and improve its outputs.

How to get started

When it comes to getting business value out of generative AI for the mining industry, there are several key considerations.

• Data security. How to maintain the confidentiality of sensitive business data particularly when using external generative AI applications like ChatGPT.
• Governance. How to establish the right policies and guardrails so employees can benefit from the technology in a secure, accurate, responsible and ethical way.
• Prioritization. It’s vital to be systematic about targeting the right business processes and specific use cases—considering both the potential impact and the ease of deployment.
• Infrastructure. There are many different routes companies can take to deploy generative AI—selecting the right one, and building the right architecture to support it, needs careful planning.
• Partnerships. Given the pace of change, and the challenges involved in developing generative AI capabilities from scratch, companies will need to consider where and how to partner with the broader ecosystem.

Like many industries, mining stands to gain hugely from this exciting technology. But to maximize the value, and minimize the risk, companies will need a well-defined strategy, clear guardrails, and responsible adoption.

Let’s work together to dig deep into the transformative potential of generative AI and unearth a wealth of efficiency, productivity and innovation for mining companies and their stakeholders.

I'd like to thank Liv Carroll , Aldo Souza, BSc Eng (Hons), MBA, FAusIMM and Marco Mayrink for their contributions to this article.

My posting reflects my own views and does not necessarily represent the views of my employer, Accenture.