Miners Say They Dig AI But the Gold Rush Hasn't Come (wsj.com)

An anonymous reader shares a report: The chairman of Barrick Gold Corp made a bold prediction in late 2017: With the help of artificial intelligence and other digital tools, the world's largest gold miner would become a technology company that just happened to be in mining. A year later, Barrick has parted ways with its chief innovation officer, chief digital officer and many of the team tasked with making this transformation a reality, according to people familiar with the matter.

The revolution in machine learning, as predicted by Barrick Chairman John Thornton and other mining executives, has yet to come. Miners have said digital technologies like artificial intelligence, or AI, will revolutionize one of the world's oldest industries in the same way it has changed other businesses, from retail to hailing a cab. Some experts say the promise of AI in mining has been overhyped and progress has been slow. Companies, including Barrick and giants such as Rio Tinto and BHP Group, are running some AI-led projects. But implementation at some companies has hit cultural hurdles. Executives haven't always engaged, projects have taken longer than expected and companies have turned to other ways to modernize operations.

Comments

That was a dumbass prediction

[Red_Forman]

You can't "A.I." your way into finding more gold.

Re:That was a dumbass prediction

[guacamole]

Not if you're mining cryptocurrencies!

Mines out here already have self driving trucks

[rsilvergun]

a buddy of mine was pretty pissed because he'd been trying to get a job driving truck at the mine for years. It (used to) pay really well. They're also using little drones to plant explosives and explore tunnels (since they can't use children anymore :) ).

I'm actually surprised there aren't more robots. I realize these aren't really AI, but from a layman's standpoint the distinction doesn't seem too important.

Re:Mines out here already have self driving trucks

[Aighearach]

They learn most of what they need to know from geological studies that are related to those explosives.

They already have good data, and well-developed statistical methods for processing it. AI here would only help on the margins, it wouldn't help them on deciding where to mine.

It would probably be useful for people who buy depleted mines and then find the nooks and crannies that have a little gold left. But the big boys have no need for it until it is so good that they don't have to hire geologists anymore.

Re:Mines out here already have self driving trucks

[frank_adrian314159]

They already have good data, and well-developed statistical methods for processing it.

Most of which was thought of as AI research back in the 80's. Schlumberger, (among many others) were bringing together advances in signal processing and "AI" to figure out where oil deposits were back then. All of this was supposedly "whiz bang" crap by the haters of the day, but these techniques (and their much better heirs) are just used today. Two morals of this story: (a) Don't be so skeptical about a field that under-delivers on its ultimate goals, but seems to spin off scads of useful technology in the process of reaching for it; (b) Those who don't remember history are doomed to repeat it.

Re:Mines out here already have self driving trucks

[Aighearach]

Yeah, but oil is way harder than gold. Gold is extracted from near the surface.

It is because it worked so good for metals that they knew to apply the computers to the harder problem of oil.

The point is, they can already easily find most of the gold in an area without much digging. Improvements only can hope to find the small amount that isn't already easy to find. Gold mines already don't mine all the ore they can find, they only mine the higher quality parts. They don't have a desire to dig up more of the lower quality stuff that they might find with an improved technique.

Whereas with oil, they can't find nearly as much as they want, and they can use almost anything they can pump out. Lots of room for improvement there.

I'm not skeptical that it might have some other use. But if there is some other use, that doesn't actually help it to be useful for the use discussed in the story. The potential existence of other uses does not in any way touch skepticism towards this application.

If it eventually replaces the geologists they currently use, it will do so without having significantly changed what gold ore is detected and dug up. Because that is not a limiting factor in gold mining.

AI isn't that smart yet.

[jellomizer]

Computers are not smart, but they are relentless. an 8 year old can add numbers together, but a computer can add numbers together for days on end, much faster then a person can, because it doesn't get board, or care about its existence, you can take the CPU fan off the CPU, and it will add numbers until it is burned out.

Today's AI systems are just big probability and correlation calculators, where the Math to do this has been around for centuries, just that with large data sets it become exponentially more complex to process. Today's computers following Moors law have grown to a point where it can handle big enough data sets over a long period of time, where it can find patterns and correlations that most humans will just get tired of doing.

The human brain is actually really good at correlations that is why when we see an Apple, we know it is an Apple, even if it is a different color, at a different location, shaped slightly differently, or even at a different orientation. However we are good at processing a 3d world, but giving big sets of 2d Data. not so much, so the computer can handle it better.

Now for Mining, our Correlation brain with experience, can give an experience minor a gut feeling on where to dig, he will be able to figure this out much faster then today's fastest computers.

We work in this space

[labnet]

Our business is one of the leaders in sensors for underground mines, so we are privy to the culture of some of the biggest players in this space.

Check this out as an example. https://www.youtube.com/watch?... as to where at least cave style mining is heading.

AI is bit a of dumb term. Automation and closed loop feedback would be a better description for what is happening in mining. For example Rio have recently completed a large automated heavy rail network in Australia. http://www.railpage.com.au/new...
Companies like http://www.petradatascience.co... are using big data to increase reliability and reduce ore dilution.

Take for example sub level mining. https://www.youtube.com/watch?...
A phd was recently presented that looked at three parameters to do with explosive density, ring spacing & tunnel spacing and dilution and fragmentation. It found varying just these three parameters had the OPPOSITE effect to what was commonly understood by industry. AI is useless unless you understand the fundamentals!

Now here where you get the big culture differences. A consultant plans the mine, a contractor builds out the infrastructure (which can be $5 Billion+), and someone else operates the mine. The operator just cares about pulling out as much ore as possible. To do real R&D that helps improve mine productivity, usually involves disruptions, which means preparing budgets years in advance and executing programs over multiple years, then trying to convince the operator to change their process based on the R&D. It takes a lot of effort, money and time. For example: to drill one 100mm hole 1km long to put sensors in, can cost one million dollars!

So yes, tech continues to ramp up in mines, with increased sensors, automation and data feedback, but the gains are slow and hard won!

It is a god rush

[houghi]

Look up what the gold rush was. Overhyped promise of money where some won and most lost. Be it money or their life.

The gold rush is, if you look into it, more about loss than gain. So it sounds like a gold rush to me.