BHP turns to AI to guide exploration

BHP is the latest of the big Australian resources companies to turn to artificial intelligence to aid in its search for untapped reserves.

Exploration and geoscience executives from the company opined this week that “with advances in artificial intelligence, the ability to apply machine learning to exploration datasets across a range of scales is exciting”.

“BHP’s technology teams have experts working in this area,” it said, without elaborating.

The move represents a continuing trend of miners trying to find better ways to sift through the huge amount of data they generate when surveying new areas as part of their exploration efforts.

Rio Tinto was a first-mover in the field when it revealed plans in early 2016 to apply big data techniques to its vast troves of exploration data. It hoped to create new joint venture opportunities by giving junior miners access to the analysis results.

Woodside Energy earlier this year revealed it is using a Google cloud platform setup to reduce the time to identify potential fields for further exploration, from over a year to just weeks.

AI’s role in the mining and resources sector in the next decade is still largely “nascent and uncertain”, Accenture said in an analysis of the sector this year.

A similar analysis by Deloitte Australia concentrates on the potential for machine learning - a subset of AI and big data. However, it looks mostly at uses in production rather than exploration.

BHP has also been working on applying machine learning in ore processing.

From 2015 through to mid-2016, BHP fed production data from its processing operations at Olympic Dam in South Australia into a CSIRO project looking at how machine learning could be used to optimise mineral processing.

The pilot project ran in CSIRO’s virtual private cloud, which backs onto AWS.

Results from the project were largely withheld; however, an abridged version of a CSIRO report was published by a South Australian government department earlier this year, revealing the machine learning use case.

It’s unclear whether BHP progressed from the pilot to a production implementation, given it was not a BHP-led project.

BHP made no mention of its artificial intelligence interests at its full year results presentation earlier today. However, the company briefly talked up its ongoing automation efforts in coal and iron ore.