The rise of data and now machine learning has made us more efficient
There is an opportunity for explorers to harvest more from data than ever before, and this is not simply because there is more of it. A shift in exploration processes has allowed the industry to create value from the step-change in data volumes generated.
The quantum of data we gather today through geological fieldwork provides us with a better-grounded baseline for analysis than ever before. However, our traditional methods of working through that data – essentially, relying on the human brain to order and then identify trends – would have been paralysed by the task.
To make the most of the vast increases in information, we have needed to introduce new processes and employ people from disciplines not previously engaged, chiefly data scientists. These professionals are specialists in building the systems to correctly receive and organise data, and work with exploration experts to apply those data through the available technologies for improved exploration results.
This new and additional technical direction within SRK is an acknowledgement that if we want to deliver the next-level results many believe possible with modern technology, things must be done differently at a process level.
The early impacts of more appropriate data management and analysis processes combined with technologies like AI have delivered outcomes many had not anticipated.
Some, if not many, had expected data-fuelled AI to dramatically change the number of targets to be chased. This doesn’t look like it will be the case, at least not in the near term or for near-surface deposits.
The results generated from well executed traditional exploration have not been radically different from targeting delineated from machine learning. This is true to the point that if computer-generated models look materially different from established models, it is the algorithms that should be first checked for errors, not the legacy findings.
This is not to say current use of data and applications of AI are not proving valuable. Far from it.
For example, what we can do today is measure the differences in absolute results rather than relying on relativity.
Exploration is a false-positive problem; that is, there are too many targets that will ultimately prove fruitless. We are today able to narrow the search by getting rid of large numbers of distracting false positives.
So, while we’re not immediately making lots of new discoveries, we’re investing more resource on the true positives.
For near-surface targeting, it is therefore about spending more time and money on the targets more likely to deliver discoveries. This efficiency should ultimately translate to an increase in the discovery rate.
At depth, it will be more about the predictive technologies unlocking prospects. A lot more predictive work will go into each hole. We are several leagues away from realising the potential of data and AI to transform deep geologies into deep mines, but those goalposts are today visible.
At the heart of the gains made to date, and the key to delivering on those future opportunities, is a preparedness to change our practices to accommodate data-driven support.
Click here to view more Future of Exploration articles and videos.
This thought leadership piece is in partnership with the Mining Journal's Future of Exploration.
