Cloud consultant and technology provider Searce believes Australian mining companies face data quality cost issues when it comes to effectively implementing AI.

Currently, many mining companies aren't ready for AI because their data is siloed and dirty, according to Searce senior cloud consultant Australia and New Zealand Andrew Kirk. Getting this foundational component right can help ensure they get on track to deal with other issues that hamper AI projects, said.

Speaking to ARN, Kirk said the industry is at an inflection point where "they see the opportunity, but with it comes the pain of exploration and experimentation". However, mining organisations are afraid to make decisions because they don't want to fail.

"The mining industry is grappling with several forces, from emissions reductions to production yield improvements," he said to ARN. "Breaking down a problem into component — some solved through process improvements, others with data analysis and others with AI — requires careful judgment.

"All this depends on having good, clean data and this foundational step is often overlooked."

While data scientists are brought in to tackle this problem, they end up being less productive because they're working with poor data, Kirk claimed.

He also said that mining companies need to determine which data sources are appropriate for their current problem and need to keep in mind that data analysis and judgments must be made about the elements of the problem.

However today, most of their data comes from tools for KPI [key performance indicator] reporting, while AI requires collecting data across the entire business, cleaning it and consolidating it.

Unfortunately, most organisations have discovered that it can be costly when they start targeting serious problems for little gain, especially if they choose the wrong path, Kirk claimed.

"We've seen this with some of our customers and it makes us step back and wonder why," he said. "I call this the 'cart before the horse' paradigm that exists in Australian mining, which is somewhat blinded by the AI frenzy that's been hyped too far."

According to Kirk, a better way to approach this is to start with the end in mind and work backward, but there are many steps to achieve that, including a small team of data scientists tuning the AI model.

"As they say, 'garbage in, garbage out', so the input data has to be of good quality and secure," he said. "This requires data engineers to set up the necessary infrastructure for data collection."

"The cost of AI might look cheap initially, but the real expense comes from preparing the data and making the right decisions on where to direct AI efforts," he said.

The operational technology areas of mining businesses usually have data engineers, but to get to the level of AI implementation, they need infrastructure adjustments.

"This is why judgment on which path to take is so important and where managed service providers come in," he added. They can help create data platforms or data lakes, consolidate the data and make it secure.

"From there, AI opportunities can be explored, as well cost-benefit analysis, which is crucial because some areas might not need advanced AI."

In May, Searce previously shared with ARN how it harnesses the full potential of AI, with it explaining how it invests into data infrastructure, establishes strong partnerships and fosters intelligent outcomes with data-driven decision-making.