Over the last few years, Artificial Intelligence (AI) has come to the foray, offering exciting use cases to boost efficiency for companies who are looking to technology to gain a competitive advantage. One company which is actively harnessing the power of AI is Botswana Diamonds, a BSE-listed diamond exploration company.
Recently, Botswana Diamonds announced that AI-powered exploration efforts had yielded promising polymetallic targets across 7,322 square kilometers, prompting it to apply for eleven prospecting licenses in Botswana to pursue these targets.
BW TechZone talked to James Campbell, managing director of Botswana Diamonds, who shared more on the exploration programme's goals, challenges, achievements and what i means for the company's Botswana operations.
Can you elaborate on how AI was used in your exploration program? What specific technologies or algorithms were applied to identify mineral deposits?
Our initial dive into the use of AI was first to understand and maximize our understanding of the data we already had. The company has acquired vast amounts of data over the decades, but this knowledge has been interpreted and compiled by many different geologists and data managers over the years.
Due to the vast array of data types, formats and vintages, it is extremely challenging and time-consuming to assimilate, clean, organize, then analyze and understand exactly what we have in the data using the traditional ‘manual hands-on’ approach. We knew this task was almost impossible for someone to do in any timely manner, so we turned to AI to overcome the challenges.
"AI" is an overarching term for the many types of computer-assisted technology, such as machine learning, knowledge reasoning semantics, and others. Our goal was to use our existing data to enable the machine to perform the complex tasks that would normally be carried out by a geologist, only with much greater efficiency and a higher level of insight. To do this, a customized AI toolbox encompassing a broad range of methodologies was used.
Was the AI system custom-built for Botswana Diamonds, or are you using an existing exploration AI platform? If so, who are your technology partners?
The answer is that both methods were used in the work. Our partner company, Planetary AI, which specializes in AI machine-assisted mineral exploration, utilized its platform, "Xplore," for the work. It's quite well established that all the ‘easy to find’ kimberlites have been found, so Planetary AI utilised a number of its existing models in the work but also developed several new techniques to analyse and infer new geological insight from our data.
How does AI improve the accuracy and efficiency of mineral exploration compared to traditional geological surveying methods?
Standard methods involve a geologist manually going through the data and extracting all the relevant information required to create compilation maps. The maps include all the important data related to the mineral deposit type of interest and are used to direct the next stages of exploration work. This work could take many months, even with a much smaller data set than we have.
The AI process enables a more streamlined approach that can handle repetitive tasks with much greater consistency thereby reducing errors. In scenarios such as ours where we have huge volumes of data the application of AI can rapidly identify patterns and anomalies that may not be achievable by human analysis. Thus the time to develop the prospectivity maps can be significantly reduced with the potential for a greater level of derived knowledge.
AI systems like the Xplore platform can increase our level of geological understanding by learning from a region's explicit data to generate and integrate new inferred data into the dataset, just like a seasoned geologist would. This enriches the data and, hence, our understanding of the geology, enabling a higher level of prospectivity mapping.
Another very important feature of AI, when applied to mineral exploration, is that it can help overcome the issues related to human bias that may sometimes skew the results.
Were there any unexpected insights or discoveries from the AI-driven analysis that wouldn’t have been possible using conventional exploration techniques?
There were certainly a number of gems hidden deep within the data that may well have been overlooked during a manual process and possibly overlooked at the time of the initial data acquisition.
How much data did you analyze for this AI-driven exploration, and what were the main sources of this data?
The Company's database consists of c.225,000 sq km of data, c.375,000 km airborne geophysical data, 606 ground geophysical surveys, c.358,000 soil sample results and c.32,000 drill hole logs combined with significant government data package. This data is a culmination of decades of exploration by multiple companies that have worked in Botswana.
Can you walk us through the process of how AI identified the 11 polymetallic targets? What patterns or signals did it detect?
In regards to the polymetallic targets, we selected a range of Xplore mineral deposit models based on known deposit types. The geodata was ingested by Xplore and analysed against these deposit models. The Xplore system produced prospectivity maps, identifying target areas based on prospectivity level.
In addition a number of additional machine learning methodologies were used to increase our knowledge of the areas. The deposit models were based on Planetary AI’s deposit models which were customised for the company and specific to Botswana. We were delighted to see that the software was able to pick up the known deposits but also identify new target areas that were in "open ground".
Now that the AI system has successfully identified new mineral targets, will you be expanding its use in other exploration areas or commodities?
The licence areas we have applied for cover a broad range of deposit types and commodities, but the work is only just getting started. We intend to use the Xplore AI system further to refine these targets towards more finite areas of interest, which will determine ground investigations and, ultimately, a drilling campaign.
Given your success with AI so far, do you foresee AI playing a bigger role in Botswana Diamonds' decision-making beyond exploration, such as in resource estimation or operational efficiency?
We have witnessed first-hand the strength and capability of AI, and we will always look to adopt it to increase accuracy and efficiency, reduce risk and improve our decision-making, from early exploration right through to resource development. We aim to maximise this potential going forward.
What challenges did you face in implementing AI for exploration, and how did you overcome them?
The challenges with many types of Machine learning is that results often cannot be explained, the so called "blackbox" approach. Using results of this type to commit to an expensive data acquisition program of an area leads to a high degree of uncertainty.
This is why we adopted a "glass box" approach in which explainable AI, based on proven geological principles, is the basis of the workflow. We combined this with machine learning to give greater insight to support the results further. The goal is that each target generated can be fully backed up by geological understanding and reasoning providing confidence to the results produced.
Do you think AI-driven mineral exploration will become the industry standard in Africa? How does Botswana Diamonds plan to stay ahead of competitors using similar technology?
Even though the imperative is to increase the global critical mineral resource base, the fundamentals of mineral exploration have not changed in Africa or globally. Historically and to this day early exploration is on the whole driven by the junior explorers. These companies are typically both resource and capital-constrained but data-rich.
This is ultimately why so many juniors fail so if new mineral resources are going to be found, then these small companies will need to find ways to better understand their data, work more efficiently, reduce risk, and ultimately enable them to make more informed decisions and focus their resources and cash on the high potential opportunities. Botswana Diamonds is evolving both in terms of our target resource targets and the way we apply technology to improve our company's overall success. We plan to continue this evolution.