The mining industry has the potential to achieve significant efficiency gains from AI. But operators must be strategic in their approach.
The mining and metals sector is being squeezed on both ends. Demand for traditional mining products continues to grow and is now joined by an increasing need for the green and rare earth minerals that will power the energy transition.
Meanwhile, on the supply side, operators face pressures from depleting deposits, rising costs, sustainability expectations and talent shortages. As a result, organisations are seeking ways to optimise operations sustainably.
For leaders, the solution to this squeeze is to become technologically curious. Smart operations like artificial intelligence (AI), digital twins, and predictive analytics are revolutionising the industry. In particular, AI’s capabilities to mine insights from data at speed have the potential to drive massive efficiency gains.
Take, for example, preventive maintenance optimisation, which typically involves labour-intensive data analysis and error-checking. Generative AI (GenAI) can automate the processing of historical work order data, enabling the quick determination of failure rates and optimisation of maintenance schedules.
In our experience, the time required to analyse work order content can be dramatically reduced from 160 hours to less than eight hours for a batch of 5000 work orders. This acceleration not only saves time but also enhances productivity, leading to operational expenditure (opex) savings of around 15–25 per cent annually.
Similarly, Deloitte has developed an executive reporting agent that allows leadership to quickly access insights from large, data-heavy reports without relying on data engineers or manual search, providing real-time intelligence for faster decision-making.
These kinds of efficiency gains cannot be ignored. Any organisation that does not act risks being left behind. But at the same time, it can be hard to bite the bullet when the implementation risks of AI seem insurmountable, such as data security concerns and uncertainty over return on investment.
Minimising AI risks
There are guardrails your organisation can put in place to minimise AI risks. Deloitte’s recent ‘AI at a crossroads: Building trust as a path to scale’ report outlines five key pillars of mature AI governance needed to achieve trustworthy AI.
The first is organisational structure. It is essential to clearly identify roles accountable for managing AI standards and addressing emerging issues. Most companies place these responsibilities in the hands of the board or C-suite, while others assign them to a centralised ethics and risk team.
In mining and metals, there’s no one-size-fits-all structure – what matters is having one. A miner operating across jurisdictions might choose a global body, while a domestic operator could entrust a C-suite member, like the chief technology officer or chief financial officer.
The geolocation of the AI platform or the server storing its data is also crucial, as some jurisdictions pose higher risks than others. This must be a key consideration for any multi-jurisdictional miner developing risk policies for AI usage.
Next are AI policies and principles – such as timelines for implementing governance standards, ethical frameworks, and risk and reporting guidelines. Having clear principles helps build employee trust in AI, particularly where there are work, health and safety implications.
The third pillar focuses on managing AI-related risks in day-to-day operations. A crucial part of good AI governance is having a system for employees to report issues or concerns.
Our research shows two in five organisations lack such a mechanism. Those with formal systems receive five times more queries and twice as many reported incidents – suggesting that those without may be unaware of emerging risks.
Mining companies often oversee decentralised operations, so a centralised reporting system is essential. Data-heavy processes, like work order analysis, could be corrupted by malicious actors feeding false data into AI models.
Elsewhere, we know of one incident where a deepfake impersonation of a key decision-maker was used to approve an otherwise unauthorised purchase. Being able to escalate AI risks quickly is key to minimising impact.
Pillar four – people and skills – is critical to ensuring trustworthy AI. By our estimate, just 56 per cent of employees in a typical organisation have the skills to use AI responsibly, emphasising the need for organisations to invest in training to boost AI skills and fluency in the workforce.
Proactive training and reskilling can reduce costs, enhance safety, attract talent, and unlock data-driven ways of working across roles. For instance, Deloitte has developed a health, safety and environment incident investigation suite that automates aspects of investigations and provides data-driven insights.
There is also potential elsewhere – mine maintenance technicians could use AI to diagnose issues, analyse data, and optimise schedules, freeing up capacity and improving safety by reducing exposure to heavy equipment.
The fifth and final pillar is adaptability – ensuring your governance systems can evolve with shifting needs and new challenges. Constantly reviewing systems and processes ensures your AI strategy remains effective and trustworthy.
It’s clear that transformative AI technologies can enable mining and metals companies to unlock significant value through productivity and workflow efficiency. Realising this value requires robust, responsible guardrails to scale the technology effectively.
By Deloitte Australia national mining and metals leader, Nicki Ivory, and Deloitte Australia partner, AI and data, Nesa Abbaspour.
This article appeared in the Winter edition of Mining Magazine.
