IBM's Super-Computer is Learning Geology
At Goldcorp's Red Lake mine in Canada, geologists are using artificial intelligence to look for valuable new pockets of ore. “What would normally take us six months is expected to take us a week.”
Watson, the super-computer owned by IBM, is brushing up on geology and is due to start spewing out drill targets at one of Canada's largest gold mines within the next six months.
Under a partnership between IBM and gold giant Goldcorp, Watson is currently being stuffed full of geological data from Goldcorp's Red Lake mine, which Watson will then process in ways humans may not have spotted. “There's lots of talk in the industry, but we're actually investing in innovation,” says Goldcorp's technical head Ivan Mullany. “Watson will be ready in the second half of the year.”
The partnership is one of the first examples of artificial intelligence getting to work in the mining industry. In a bull market, the temptation is to just “buy another truck” to boost output, Mullany says. But with gold prices down from their peak in 2011, the industry's largest players have been looking in earnest for new ways to squeeze more from what they already have.
Goldcorp and IBM have a joint team of ten staff working on Watson, digitising 80 years of mining data from Red Lake, a vast underground complex that has produced nearly 30 million ounces of gold since it opened in the 1940s. Conventional computers can typically process two mine models at once, but Watson is being fed
150, plus multiple layers of additional data, including maps, surveys and thousands of drill samples.
Goldcorp's geologists have now started firing questions at the super-computer, looking for new drill targets and left-field correlations, pushing Red Lake's mining team into new pockets of high-grade ore. “Within an hour or two, you can ask a hundred questions,” Mullany says. “What would normally take us six months is expected to take us a week.”
The prize is potentially a big one: Goldcorp's exploration budget for 2018 is $125m, so any improvement in the overall quality of its drill targets could substantially boost group returns. In the oil industry, by rethinking human approaches to geology, Watson's partners have lifted revenue to the tune of hundreds of millions of dollars, IBM says.
Artificial intelligence, which is ideally suited to data-heavy applications, from medical diagnosis to weather forecasting, could find dozens of other uses in the mining industry, coming up with new mine-plans and maintenance schedules, or trawling through government data to find new geological basins. “If we can get this to work in exploration, it's pretty clear it can work in all the environments,” Mullany says. “I guess people are waiting to see what happens.”
Royal Gold's founder proposes radical new way to rank mines & industrial assets
First Quantum to Nevsun:
six top targets for Lundin's copper company, Lundin Mining
Plus: China is mopping-up cobalt and lithium, Glencore's response to trade tariffs, out and about with Ross Beaty
Fund manager Julian Treger, the driest man in London, answers business questions from readers