How a driver’s licence and business card help explain artificial intelligence in mineral exploration

The terms ‘artificial intelligence’ (AI) and ‘machine learning’ (ML) conjure up vastly different images in people’s minds. For those who do not work closely with these technologies, they can be difficult concepts to understand, and people are less likely to accept technologies and processes they do not understand into their work and home lives.

Sam Cantor, Section Head, Economic Geology, Minerva Intelligence, decided to prepare a short talk to help explain these complex concepts and how they relate to mining and exploration at AME’s new MinEx Talks event held in downtown Vancouver in April 2019. The TED Talk-style format appealed to him and he had a story to tell.

“Geologists are storytellers by nature,” said Cantor, “Anyone who has been in the field knows that. MinEx Talks gave people who already like to do that a more official platform.”

For several months before the MinEx Talks event, Cantor had been using his driver’s licence and business card to help explain the roles that perception and cognition play in an AI system.

“The first time I explained it that way,” said Cantor, “I just pulled out my wallet and I had my business card on the table. I showed them the difference between my ID and the business card, and I did notice that for people who didn’t get the concept, this example rang a lot of bells.” He workshopped the analogy with his colleagues at Minerva Intelligence over several lunchtimes to refine his explanation.

During his 12-minute MinEx Talk presentation, Cantor explained that an AI system involves both perception and cognition, working together. Perception identifies the immediately obvious information. For example, people know instantly what a driver’s licence is by its appearance. Other observations require an extra cognitive step, for example the word ‘location’ might not appear in either the licence or the business card, but with a working knowledge model of how humans organize land into countries and cities, people are able to recognise that ‘BC’ is a location, specifically a province of Canada.

“Perception happens before you can even think about what a business card or driver’s license is.” said Cantor, “The cognition stage happens when you need to take a couple of seconds and really think about what it means.”

AI and ML have been successful in mining when applied primarily to problems of perception, said Cantor. Preventative maintenance on haul trucks is one example. There are mostly concrete variables involved, such as temperature and wear, and success is easily measured.  Trying to identify a promising mineral target during an exploration program is much more complex with so many variables and too little data. But, although AI can not always generate targets with high confidence, it can be applied in other areas of exploration geology.

“What you want it to do is intelligently and reliably take care of all the other exploration activities that you have to do as a geologist,” said Cantor, “So it then allows you to spend more time on the actual cognitive tasks you are trained for.” Digitizing maps and data or organising GIS layers are examples.

“Any AI process that reliably returns time to the exploration geologist and augments their capabilities is a good one,” said Cantor, “Anything that cuts down bias gaps, or cuts down on tedium or any tasks that eat up our time, that’s where I think AI is really, really helpful.”

Cantor received positive feedback after his MinEx Talk and was pleased to have helped some people in the audience better understand AI and ML. Preparing for the talk also cemented some ideas in his own mind and helped him prepare for other presentations.

“Nothing cements a concept in your head quite like giving a simplified talk,” said Cantor.

Watch Sam’s AME Minex Talk.

If you are interested in participating in AME MinEx Talks 2020, check out the MinEx Talk Criteria and submit your abstract.

By Kylie Williams

Photo Credit: @ronniejphoto