Article
When is AI right for you? When is it wrong? And how are you to tell?
September 8, 2025
AI is becoming a commodity, and the available versions are soaring in number. But not all AI is the same. Fortunately, there is a way explorers and investors in the minerals industry can identify the solution – and it might be just one – that will cut through their workload and speed up decision making. It’s just a matter of finding the right AI mindset.
There are presently two main types of Artificial Intelligence – machine learning and Large Language Models (LLMs). Simply put, machine learning is about calculation and LLMs are about identification – extracting insights from text documents, trends, and so on.
With machine learning, the necessary algorithms are largely free and not hard to use, which is why it’s almost become too easy to put some AI systems in place. The trouble here is that this blanket application of AI can often produce results that make zero sense, because the outputs you get will depend on the data the AI was trained on. If that data has no relevance to your industry, or if it exhibits no underlying understanding of the science within your sector, then there’s little chance that what comes out will actually be of any use to you.
Is too much AI a bad thing?
This commodification of AI is an advantage because it lowers the barriers to entry on the technology side. But it becomes a positive disadvantage if the training data cannot keep up with the real-world tasks or scientific areas it’s expected to analyse. AI projects fail, companies become disillusioned, and major benefits may be lost.
Similarly, many of the commercially available Large Language Models are not fit for the job. They’re not specialised, and they don’t deliver consistency.
What you need to do is build a system that is resilient and takes the best and most suitable characteristics from all the models out there and combines them with an understanding of your customers’ problems.
You have to make sure the AI is tailored to their challenges, their science and their data; a solution that will actually get better every time you give it a mining project to analyse.
Only a precisely trained and tailored AI can help an industry as complex as mineral exploration
Another way to think of it is to consider what you would expect of an intern you hired straight from university. They’re good, but you wouldn’t give them the work you’d give to someone who has 15 years’ experience behind them.
Most LLMs are effectively AI interns. They’re extremely smart, they’re extremely skilled, but they’re still ‘youths’ in that sense. What we do is give those AI interns a Masters in geoscience and risk analysis. We give them the perspective of someone who’s been doing the job for a decade. We give them the experience of maybe 10 years’ worth of evaluating mining projects.
We bring them up to speed, and we make them relevant to the field they’re working in. If you don’t take that vital step; if you rely on an off-the-shelf LLM to make your decisions for you, you’re basically letting the intern run the business….
How AI can cut analysis time from weeks to days for investment teams studying new proposals
Geologists who are coming to the ends of their careers now presided over an era when finding mineral deposits wasn’t actually easy… but it did have far fewer complications than it does today. Deposits were generally closer to the surface, discoveries were more frequent, it was easier to put a project or mine in place, and there were fewer constraints on supply chains. All these issues have become tougher today, and there’s the additional question of sustainability to contend with. Consequently, the industry requires a wider range of skill sets, and a broader base of expertise in geology, data and business.
But this means that the world is changing for investors too. They need to be able to assess the value of new proposals coming their way more rapidly than ever, and with more accuracy than ever. If mineral exploration cannot lift its average success rate of just 5% for potential new deposits, then hitting the necessary targets to achieve net-zero by 2040, aid sustainability and ensure security of supply for critical minerals will slip ever further out of reach.
Extracting the key facts from 300 to 400-page proposals, not to mention unstructured data rooms and decades of map data and reports, has traditionally taken weeks or months in order to create the thorough analysis that, say, feasibility or due diligence studies require. We need to shorten that to days and for simple screening, this should now be minutes.
Investors must make sense of complex or contradictory data shaped for geologists rather for financiers, yet do it without missing key information that might be hiding in the detail. And they have to be confident that any appraisal delivers an accurate view of a deposit’s potential, not one clouded by over-enthusiastic marketing from the exploration company.
Fortunately, technology is accelerating the ability to leverage and comprehend giant sets of data in ways impossible in the past (or for that matter, even a year ago). The enormous strides in AI are bringing astonishing levels of insight to the desks of exploration investment teams. Importantly, it’s insight that – finally - respects and understands the complexity of the underlying science. New, appropriately trained models can recognise the critical facts, not be swayed by the wrong ones, and produce summaries that crystallise the vital content, without overlooking significant details, even if hidden deep in the data. They seek, they find, and they don’t miss out.
While it might be an exaggeration to call it a ‘superpower’ – and there is no shortage of hype around AI that must be treated with caution – it is nonetheless true that the possibilities for rapid AI-powered analysis of mining proposals puts investment companies in a stronger position to de-risk decisions than they have ever been.
Interested in exploring what the right AI mindset could do for you? Reach out and book a walkthrough with MinersAI here.