• MINING INDABA
a key part of mining professional skills development . A range of professions could include : technicians ; engineers ; and applied scientists . Obviously , with each progressing level , the expertise deepens and becomes more abstract .
The capability of the typical mining graduate in the domain of digital computation depends on the discipline and the extent of quantitative training they receive . Generally , their capability in this realm is deficient for some of the disciplines . This domain breaks down into in-discipline mastery and more generalpurpose numerical computation and analytics ( e . g ., Fig . 1 ). The general ability to visualise , adequately manage and process data , automate trivial or repetitive workflows and diagnose software and hardware issues are generally not systematically taught at the tertiary level . More advanced techniques such as statistical or dynamic modelling , as well as analytical instrumentation and control are typically missing as well . These techniques would be needed to ( non-exhaustively ): specify , engineer and control remote machinery ; engineer , troubleshoot and use digital twins ; analyse operational and business processes ; and study and understand complex natural phenomena through data / information fusion ( e . g ., prospectivity mapping and deposit formation ). Equivalently , relative to probable future needs in the mining industry , the current education overemphasises traditional observational methods at the expense of theoretical and computational training . As more observations become non-contact , there would be more requirement for integrated data management , multidisciplinary data fusion and modelling , visualisation and synthesis .
Trans-disciplinary techniques are generally not taught directly within mining disciplines at any level , despite its popularity in the adjacent fields . The rise of artificial intelligence techniques such as machine learning are driven by the increasing desire to extract value from high-dimensional , variably structured and often voluminous datasets , as well as automation . This desire is somewhat heightened by the fact that mining resources data is expensive per sample and many legacy datasets can never be replaced ( e . g ., exhaustion of deposits ). In this case , with expertise turnover , the only remaining knowledge is contained within the data . Trans-disciplinary education is probably the most difficult to accomplish after the tertiary level , because of the very broad and fundamental pre-requisites that cannot be generally met outside of the tertiary environment ( e . g ., on-thejob training ). For instance , machine learning requires substantial undergraduate training in statistics , mathematics , computer programming , optimisation and basic computer science . Similarly , mastery of physical automation is predicated on the knowledge of control systems , computer vision and navigation . It is obvious that in the future where data and sensors are more abundant in the mining industry , there would be an even greater demand for specialists who can leverage transdisciplinary techniques .
An implication of this crosswalk of the state of mining expertise and potential future mining conditions reveals that typical mining education today is unlikely to be sufficient for future mining needs . This creates an effective and still-enlarging skills gap , where from
36 • African Mining • January 2023 www . africanmining . co . za