How the technology of tomorrow is helping Saskatchewan consumers and businesses today
If you think Artificial Intelligence (AI) and Machine Learning (ML) are science-fiction concepts that exist only in Hollywood movies, like The Terminator, Blade Runner or Ex Machina, think again.
Not only are AI and ML becoming more mainstream and ubiquitous in our daily lives, they’re also behind the scenes working to make consumers and businesses more successful, right here in Saskatchewan.
In fact, AI and ML technology is helping a mining company increase its efficiency and reduce waste and budget-conscious consumers to save a few bucks on their weekly grocery bill!
So what exactly are AI and ML and what role are they playing in the everyday world of businesses and consumers in the 21st century?
The concept of Artificial Intelligence has actually been around for centuries as the ancients dreamed of and told stories about machines that would do the work—physically and mentally—of human beings. The first “computers’’ developed in the 19th and early 20th century were glorified calculators able to make complex calculations, but not much else.
Over the last 60 or 70 years, with the advent of transistors, the silicon chip and computer languages, AI evolved to become the science of building “intelligent machines” that think—reason, weigh options, make decisions—just like humans. In the late 1950s and early 1960s, scientists theorized that, rather than programming computers with information, they should teach them to think for themselves.
With the advent of the internet, and its massive stores of digital information, computers could make the transition from giant calculators to “thinking machines.’’ Thus AI gave rise to Machine Learning—which has driven the proliferation of devices that help us in our daily lives.
As one expert put it, “while machine learning is based on the idea that machines should be able to learn and adapt through experience, AI refers to the broader idea that machines can execute tasks ‘smartly’.’’
Today, we use AI and ML when seeking information (Google), giving commands (Siri, Alexa), riding in self-driving cars (Tesla), playing computerized chess and other games. When planning a business trip or vacation, our personal device will automatically offer advice on weather and travel alerts. Everything from birthday parties to marketing campaigns can be planned and executed with help from your AI assistant. This very story and the publication it appears in has been aided (immeasurably) by AI and ML through spell checks, internet searches, computer pagination and word processing software. Of course, these are now common applications of AI and ML we see in our daily lives. But what kinds of applications of AI and ML will we see in the future?
Mining
Over at Saskatchewan Polytechnic’s Digital Integration Centre of Excellence (DICE) in Saskatoon, the province’s only Technology Access Centre (TAC) has been working for two years with Cameco Corp. on applying AI and ML to improve the accuracy and efficiency of uranium mining in Northern Saskatchewan.
Because of the extremely rich and highly radioactive uranium orebodies found in Saskatchewan, like Cigar Lake and McArthur River (the world’s two richest uranium orebodies), the uranium must be mined remotely by machines using a technique, known as jet boring. By using freezing brine to stabilize the ore and surrounding rock, jet boring allows the orebody to be mined safely. “Our high-grade orebody is relatively small by total volume compared to other commodities, such as iron ore or potash, so our mining techniques require higher precision,’’ says Jeremy Breker, Cameco’s vice-president of technical services.
Breker says mining a high-grade uranium orebody requires “a carefully designed set of parameters that are fed into the jet boring machine to extract only what is economical.” Those parameters take into account the ore grade, jet bore machine pump pressure, rates of flow, etc. in order to excavate the optimal cavity shape. “There is an incredible amount of data to analyze and, while our engineers are very competent and efficient, it is impossible to consider all data using conventional analytical techniques,’’ Breker says.
That’s why Cameco and Sask Polytech are using AI/ML to program the jet boring machine based on a “multitude of data sets’’ to enable Cameco’s mining engineers to develop a more accurate jetting “recipe’’ that improves mining efficiency and reduces waste. “AI/ML technology improves the precision of our mining operations, which reduces our operating costs. It can improve the efficiency of our human resources and has the potential to optimize our use of energy that drives the mining machine,’’ Breker said.
Dr. Terry Peckham, Sask Polytech’s DICE TAC director, says the AI/ML research project with Cameco is the first of its kind undertaken by a polytech for the mining industry in Canada. “This (AI/ML-assisted mining) is not a process that’s commonly done in the mining industry. That’s partly because of the geology they have in the area,’’ Peckham says. “This is a completely unique system.’’
Peckham says using AI/ML helps Cameco achieve a number of goals, including “producing as little waste as possible,’’ while leaving a mining cavity that is “more predictable in size and more stable,’’ reducing the likelihood of a cave-in before waste rock is backfilled into the cavity.
But the big advantage of AI/ML is to bring more data into the mining process in less time than would be economically or practically feasible otherwise. “We’ve spent a couple of years working on the mining machine algorithm. We took a look at all of the data they used. It took them a long time to come up with the right recipe for each cavity they drilled,’’ Peckham says. For example, Breker estimates it takes four to five hours to develop a “recipe” for drilling a portion of the orebody, while using AI/ML technology would typically reduce that time by an hour.
While AI/ML will reduce the time it takes to drill the orebody, Peckham says it won’t reduce the number of engineers and technicians working on a drilling project. “Everybody thinks AI is going to take all the jobs away. Actually, what this has done is given the people that create these recipes extra information. So it’s a hybrid system. We analyze data a lot faster. And they’re able to look at that and use it to modify the recipe.’’
While the AI/ML research project is ongoing, Peckham believes some progress has been made improving the mining process at Cigar Lake. “There are actually some very positive gains that have come from this (AI/ML research project).” While Peckham wasn’t able to be more specific, he says “it has improved their mining processes.’’
Breker agreed that the AI/ML research project with Sask Polytech has shown potential to improve the efficiency and accuracy of Cameco’s mining techniques. “While we have made strides with the AI/ML tool, we think the tool can be further optimized to enhance the unique jetting system at Cigar Lake.”
“There are likely other applications for AI and ML at other operations and we will likely explore those as we gain experience working with the technology and service providers such as Sask Polytech’s DICE,” he adds.
Food
While AI and ML technology can help big companies increase their productivity and profitability, what can it do to help average consumers and small businesses make their dollars go further?
Well, how about an app on your smartphone that would help you save money when buying groceries? Melanie Morrison, a University of Saskatchewan psychology professor, noticed that many people are struggling to afford groceries, especially now when food inflation is running at 10 per cent.
In fact, Morrison admitted she had trouble staying within her food budget when she became a single parent. “I thought, if I’m having concerns about this—making smart purchasing decisions—then I bet other people are as a well.’’ She also noticed the “absence of pricing data anywhere.’’ So about four years ago, Morrison created BetterCart, a free app that allows people to compare the price of groceries at major grocery stores in their area.
Lacking a technical background in software development, she reached out to Stephen Parslow, a “serial IT entrepreneur’’ based in Ontario. “We share similar values, we communicate well and we wrapped ourselves around this problem,’’ Morrison says.
The BetterCart app uses AI and ML to assemble a huge amount of data on pricing on about 60,000 different items per store and compares those prices to the same items in other stores in your area up to a 50 km radius. And all of this information is literally, instantly at your fingertips at no cost.
“Doing this sort of cost comparison manually would be incredibly cost and time-prohibitive,’’ Parslow says. “Each week, we process over 30 million new price records from over 3,500 stores across 29 different grocery and drug store chains.’’
Five or 10 years ago, Parslow says this amount of data would not be readily available, but, with the pandemic and the rise in online shopping, most, if not all, prices are now publicly accessible. Figuring out how to do price comparisons of similar products with different brand names was an added challenge.
“We had to develop AI technology to allow us to do product comparisons between stores. So when we’re looking at comparing something between Safeway and Sobeys and Superstore, we need to be able to know that two-litre skim milk is the same one (as in the other stores).’’
However, aside from selling advertising on BetterCart, it’s hard to make money developing and distributing a free app. So Morrison and Parslow also formed BetterCart Analytics to monetize the data collected through the AI technology. “Our primary focus right now is on our business analytics suite,’’ Parslow says. “We’re hoping to use capital that we earn working in our business analytics to help fund our consumer effort.’’
Morrison says the pricing data compiled by the BetterCart app is also highly valuable to food producers, distributors, retailers and other participants in the food industry. “We have five billion product records in our data store. So we have a lot to offer.’’
And this is just the beginning of BetterCart’s plans for market domination in the area of food pricing. “Because we’ve been collecting data for several years now, we’re sitting on a wonderful stockpile of historical pricing information,’’ Parslow says. “We’re just scratching the surface.’’
“We’re working on a lot of really cool technology like being able to detect regular versus sale prices, being able to determine when a product is going to go on sale, based on historical trends. We’re excited to be getting into things like predictive price analytics.”
Morrison says BetterCart can also help people with dietary restrictions and provide information on health food products, natural or organic products, and gluten-free products. “There isn’t a lot of pricing insights around these, not only for consumers, but for producers of these products.’’
Morrison believes BetterCart has the potential to dominate the Canadian market as the ‘go-to’ source for food pricing information all the way up the food supply chain. “We want to grow exponentially,’’ Morrison says. “We expect to be over $10 million in annual revenues in five years. Along with that, we want to be hiring new employees.’’
BetterCart hired its first employee in 2019, which has since a grown to a total of seven by late 2022. Morrison hopes to have 20 employees in two to three years and 50 in 10 years. Morrison believes the sky is the limit for BetterCart to grow in Saskatchewan.
“There are so many untapped opportunities right now. It’s really exciting times in the food industry in Canada.’’
