The technology of the future is being created in Canada today. The Canadian government is embracing artificial intelligence (AI) research, investing $125-million earlier this year to launch the Pan-Canadian Artificial Intelligence Research Strategy.

“By encouraging cutting-edge technology like AI while at the same time creating a culture of lifelong learning, we will be with Canadians every step of the way as they lead us into a future filled with possibility,” Minister of Finance Bill Morneau said in a statement last March.

While sci-fi movies and literature envision sentient robots as the primary application of AI, the truth is a little less scary and a lot more useful. Instead of producing human-like machines capable of world domination, the government funds will be used to develop machine learning in a variety of practical areas such as data science and health information.

One fear surrounding AI is job loss, but evidence is proving it’s the other way around.

Out of 993 companies surveyed currently using AI, 83 per cent said it’s creating jobs within the business.

Gartner, a research firm, revealed AI would create 2.3 million jobs while eliminating 1.8 million by 2020

The University of Alberta’s machine learning lab received $25-million in funding thanks to the federal government’s initiative. The lab has been at the forefront of AI research in Canada for decades already.

As a result, Google has set up shop in Edmonton with DeepMind Alberta, the company’s first international research base outside of the U.K. Partnering with U of A researchers, the company is focused on developing complex machine learning algorithms.

Whether it’s teaching bots how to play complex games like StarCraft II or using machine learning to diagnose life-threatening eye conditions, DeepMind utilizes more than 400 employees worldwide to make breakthrough AI discoveries. The company has already opened another research lab in Montreal.

We are already interacting with AI on a daily basis through Siri, targeted ads, Nest thermostats and suggested content on platforms like Netflix.

However, as technology advances, AI is going to become further integrated into our lives, revolutionizing the automobile industry and potentially the health care system. It will create more jobs than it eliminates in the form of trainers and those who explain the decisions AI systems make.Apple’s “digital assistant” Siri is one way we already interact with AI on a daily basis. Siri recognizes speech and can do a variety of tasks, such as booking an event on your calendar or calling a contact on your phone. Photo by Paul McAleer

According to Statista, the global AI market is currently a $2.4-billion (USD) industry. Next year it will be worth $4-billion. By 2025 it will be worth nearly $60-billion.

Alberta is known for its oil and gas industry, but investing in AI could prove just as profitable in the future.

In addition to its rich academic history in the field of machine learning, Edmonton is also heavily involved with self-driving cars, planning to test an AI-powered vehicle at the U of A.

Calgary is not one to let Edmonton go unchallenged, whether it’s with hockey or AI research. A number of local businesses and institutions are utilizing machine learning in unique ways. 

Despite these advances in AI technology there are still many questions about the industry revolving around definitions, ethics and dangers. Turning to Siri is one way to answer these questions, but then again, is she a trustworthy source?

Key terms

  • Algorithm: A set of rules or procedure a computer follows to reach an objective or solve a problem. There can be multiple types of algorithms. For example, a search algorithm retrieves relevant information using variables the user inputs
  • Machine learning: An application of AI to work outside of the parameters of its code for the purpose of learning independently. There are multiple types of machine learning used to analyze large amounts of data.
  • Data Science: A field that combines statistics, computer science, applied mathematics, and visualization to analyze and contextualize mass sums of data.

Researcher talks hybrid intelligence

Guenther Ruhe is a University of Calgary professor who leads the Laboratory for Software Engineering Decision Support (SEDS). Decision support systems often involve machine learning elements, but they work best when paired with human explanation. Photo by Paul McAleer

Originally from Germany, Guenther Ruhe leads the Laboratory for Software Engineering Decision Support (SEDS) at the U of C.

He became interested in AI in the 80’s when the idea was at a conceptual level. As technology advanced, he realized AI would have a huge role to play in software engineering, a discipline that involves millions upon millions of complex lines of code.

Experts Define AI
“Artificial intelligence is the umbrella term, subsuming all the effort to use machine intelligence for reasoning, planning, prediction, to classify objects based on characteristics and doing all the intelligent searching to find the best strategies in different contexts.” — Guenther Ruhe, University of Calgary

“The more you use additional techniques, referring to classification, to machine learning, to predicting, to planning — the more you do of that, the more you move into AI,” he explained.

An author of several books and more than 200 peer-reviewed research papers, Ruhe’s interests include product release planning, decision support systems, and pretty much anything involving an extensive amount of data.

In addition to his extensive portfolio, Ruhe is focused on bridging the gap between academic research and practical results that can solve societal issues.

“In research, there’s often the perception that the more papers you write the better, and the more you get referenced the better,” he said. “This is nice, but having an impact on society is another aspect.”

During the aftermath of the Fort McMurray wildfires, Ruhe and his team compiled data from Twitter to see if current apps were meeting the needs of actual victims.

“We analyzed more than 70,000 tweets and found out what are the actual needs of people and designed apps which intended to better serve their needs,” he said. “The most surprising thing was that in all the apps existing for wildfire emergencies worldwide, the top ten needs are not contained in any of these apps.”

Is AI dangerous?
“There’s always risk that it can be misused. On the other hand, I argue very much to extend the usage and applicability of AI, but making it carefully and consciously.” —  Guenther Ruhe, University of Calgary

These features included fire alarm notifications, food and water requests, safety guidelines, emergency text messages, locations of gas stations, medical centres and emergency zone maps.

The research utilized methods of machine learning through classification and textual analysis.

When it comes to product releases and decision support systems, Ruhe collects data in a few main areas such as timing, functionality, quality and features.

“You can map this into comprehensive and complex search algorithms,” he said. “And then running optimization to make offers what, from a user perspective, might be the best release for the product.”

Even with complex algorithms crunching the numbers, the human component is still a key part of extrapolating useful results.

“I would not make a sharp boundary between machine and human intelligence,” Ruhe said. “They’re not competing with each other, but they are complimenting each other. The real trick is to find hybrid intelligence, understanding that humans can do things better than the machine.”

Ruhe added decision support systems are a great example of hybrid intelligence.

“The idea is not to replace a human by any means, but to add to the competence of the human expert to make decisions in situations where things are not clearly defined,” he said.

Even though technology is capable of providing excellent solutions, Ruhe explained that it’s important to understand the process and reasoning behind the results.

“The more sophisticated things get, the more difficult it is to convince people to use them,” he said.

Ruhe believes AI has the potential to improve many different industries in Alberta, but the heavy focus on oil and gas is a hurdle to leap over.

“Companies need to not only see the potential, but they need to take the risk to bring (AI) into their process, into their development,” Ruhe said. “It does not come for free, they need to work with researchers.”

On the other hand, Ruhe believes it’s important to encourage students to explore research with technology such as AI.

“We need to educate the next generation of students much more towards that,” he said. “The full power of AI comes from the combining of other disciplines.”

MyHEAT addresses energy efficiency

Jeff Taylor, co-founder of MyHEAT, believes his start-up can help homeowners save money and energy. By using infrared cameras and airplanes, MyHEAT can produce thermal images of entire cities to show where structures are losing heat. Photo by Paul McAleer.

On the other end of the business spectrum, there are a number of start-ups in Calgary utilizing machine learning in creative ways.

Publically launched in 2015, MyHEAT uses airplanes and advanced thermal infrared cameras to generate heat maps of entire cities. From there, the team at MyHEAT can look at individual structures, seeing exactly where houses and commercial buildings are losing heat.

Experts Define AI
“A well-crafted algorithm tries a bunch of different things and once it reaches the end, it doesn’t know where to go. “[Artificial intelligence] is more thinking than just following a bunch of rules — as you provide it with more information, it learns.” — Jeff Taylor, MyHEAT

CEO Jeff Taylor co-founded MyHEAT with Dr. Geoff Hay from the University of Calgary.

“Dr. Geoff came home to his new house. It was colder than he expected and he thought, ‘Man, I just wish I could go to Google Maps and see where heat was escaping out of my home,’” Taylor said.

Specializing in geography and remote sensing, Hay embarked on his quest and successfully launched the U of C HEAT pilot project, which mapped out approximately 38,000 Calgary homes. The project won the Massachusetts Institute of Technology’s (MIT) grand prize for global climate change solutions.

Taylor was excited to partner with Hay because his idea was unique, especially for a city like Calgary. “It’s a very much an oil and gas city,” he said. “Here’s something that’s related to energy, but related to conserving energy and making the world a better place.”

Taylor believes they’re making an invisible problem visible, incentivizing homeowners to take action when they can see where their house is losing heat.

“Thermal imagery looks like a big black and white photo,” Taylor said. “I can see the whole city, but now I have to extract out every single building.”

It took two days to map out Calgary. The weather changed during the timeframe, so the team had to normalize the data while accounting for microclimatic factors like varying elevations and winds.

Is AI Dangerous?
“I do think it’s a concern when you start looking at a robot that has artificial intelligence. I think in general, the vast majority of developers are doing it for positive reasons, but there are developers that can develop things for nefarious reasons. Everything can be hacked. We have to be really smart about the rules and safeguards in place to protect us.” — Jeff Taylor, MyHEAT

MyHEAT uses advanced algorithms to assess the data generated from the thermal images. Each house is assigned a HEAT score depending on how much energy is being lost. Generally, heat loss is commonly seen near windows, doors, attic hatches and walls connected to garages.

“Right now we have an algorithm that looks at the house and assesses what we believe is general heat loss,” Taylor said. “The computer doesn’t know elevation or orientation. There’s a bunch of other things it needs to learn and interpret. It doesn’t magically know, you have to help train it.”

With so many factors affecting heat loss, MyHEAT’s use of AI is a work-in-progress helmed by two of the company’s five employees.

“We’re using artificial intelligence to a certain extent now, but I wouldn’t say we’re cutting edge yet,” Taylor said. “From a start-up perspective, it comes down to a balance of resources, so how many people can we hire to do that sort of development versus how many people we need to deliver a product out the door.”

Each home is assigned a HEAT score. The lower the score, the better, because the house is conserving more energy. Photo by Paul McAleerDespite being a start-up, MyHEAT already has their sights on some of the industry’s biggest companies.

“The main way we generate revenue right now is that we’re highly focused on energy utility companies,” Taylor said.

“Utilities spend 9-billion dollars a year on energy efficiency programs and we feel like that’s a real good fit for the data we have.”

Taylor explained that the majority of utility companies actually want to lower rates of energy consumption. “It seems counter-intuitive because they make money selling energy, but they are required and incentivised, often by their regulatory bodies, to do that,” he said. “Energy efficiency is the cheapest form of energy production.”

However, Alberta is still in its infancy when it comes to offering the same level of incentives as other provinces. Taylor said the newly formed Energy Efficiency Alberta is a step in the right direction.

As for the future of MyHEAT, Taylor said they are planning to continue to develop its algorithm and map out other cities around Canada and the world.

“Our goal in Canada is to be able to, for the price of a stamp or two, provide the homeowner a high-res thermal image of their house and how they compare with their neighbourhood ratings,” he said.

 The not-for-profit Cybera makes sense of mass data

From left to right, data scientist Byron Chu, vice president of technology Barton Satchwill (on-screen), director of communications Meagan Hampel, and data scientist David Chan, all believe AI has a promising future in Alberta. Photo by Paul McAleer

Cybera is a not-for-profit organization primarily focused on providing high-speed networks and computing resources across educational institutions in Alberta, but they also work with researchers, entrepreneurs and other non-profit groups.

Rather than inventing new technologies, Cybera monitors breakthroughs and figures out if they are applicable to the public sector.

“Our role has always been trying to take useful new technology and act as a catalyst to help other people adopt it,” said Barton Satchwill, vice president of technology.

The government-funded organization was one of the first adopters of cloud computing and now they have their eyes set on data science, a field heavily rooted in machine learning algorithms and data crunching.

While Cybera has been looking into data science since 2012, the five member team that includes Satchwill, Byron Chu and David Chan was officially formed last year.

Experts Define AI
“There’s the science fiction idea of artificial intelligence, where we have sentient machines and what not. That’s called general artificial intelligence, that’s what Stephen Hawking and Elon Musk are very concerned about — and probably rightly so — but that is still some ways in the future I think.” — Barton Satchwill, Cybera

As for projects, the data science team has investigated Internet speeds across Canada, revealing how download speeds are increasing at a higher rate than upload speeds. Their research revealed that New Brunswick had the fastest Internet speed out of all the provinces while Alberta ranked 10th.

The team can’t disclose any details about current projects but Meagan Hampel, director of communications, said data science is also proving useful for educational institutions.

“School boards can come to us and say that they’re noticing a lot of students who aren’t attending and missing classes, and then ask us to run the data,’” she said. “You can look at weather patterns or socioeconomic factors that are going on.”

Another data science team in Chicago looked at post-secondary graduation rates, analyzing factors such as age and prior education.

“It’s important to note that a lot of this stuff isn’t new. Applying some of the machine learning techniques that are coming out is the new angle to it — to better identify and better predict some of these factors,” Chu said. “That was a big difference of what was done in the past.”

Satchwill believes the ability to analyze massive sets of data is useful in a number of different areas. In one medical example, he said researchers compared genetic essays of cancerous tissue versus healthy tissue, discovering one affected gene was associated with insulin production. A diabetic drug turned out to be the best form of treatment for the patient.

“With huge sets of data, you can find out what the actual results really are, not just what you think the results are or predict them to be,” Satchwill said.

With offices in Calgary and Edmonton, Cybera believes the future of AI in Alberta looks promising, and they anticipate data science teams to pop up in a variety of different industries.

“A lot of momentum was already built up [five years ago], but it was coming out of areas like Silicon Valley and New York City,” Chan said. “There wasn’t much in Alberta at that time, but over the years, you’ve seen the adoption increase here as well.”

DeepMind chose Edmonton for its first international lab because of the U of A’s strong machine learning department, and Satchwill believes other brilliant minds will continue to flock to the city.

Is AI Dangerous?
“There’s a lot of fear and anxiety about job loss or about how machine intelligence is dehumanizing. I’m not so worried about that. I’m more inline with augmented intelligence. If people think more about their relationship with their phone and Siri, I think that’s more like what AI will become in the future.” – Barton Satchwill, Cybera

“People around the world will be coming to Edmonton to be part of this environment that’s growing up around here,” he said.

As for Calgary, the Cybera team believes there are lots of opportunities for start-ups to take advantage of AI and the municipal government is embracing data science, but it would be hard to replicate U of A’s machine learning culture they’ve reinforced for decades.

While Satchwill can’t predict what role AI is going to play in our lives 10 years from now, he’s noticed a trend with the adoption of new technology.

“The funny thing about technology is as it matures, as it gets better, it tends to disappear,” he said. “Nobody really thinks about the technology that’s in the phone nowadays, but it’s staggering if you stop to think about it.”

From Siri to spam filters, we are already interacting with AI on a daily basis without being consciously aware of it.

City transparency opens doors for data scientists

With the launch of the City of Calgary’s upgraded Open Data Portal in Nov. 2016, the city set itself apart by presenting datasets in creative ways.

Rather than just offering raw data to download, the improved Open Data Portal allows users to interact with graphs, charts and maps.

“We felt like it would be important for people to be able to visualize the data, whether through maps or other types of information before they decided whether they wanted to download it,” said Open Data Strategist Janusz Gawor.The City of Calgary’s Open Data portal is visualizing data in innovative ways. This heat map updates frequently, showing the prominence of traffic collisions in different areas across the city. Screenshot from the City of Calgary’s Open Data portal

One of the most popular datasets is an interactive heat map for traffic incidents in the city, shining light on the city’s problem areas and even allowing users to filter results such as incidents involving pedestrians.

The open data initiative is not limited to Calgary, but it’s part of larger movement across Canada.

“It’s an international movement to open up datasets,” Gawor said. “For the interest of transparency and accountability, governments are saying citizens should have access to that data.”

Other datasets look at economic and environmental factors, measuring how the data stacks up to milestones set by the City of Calgary.

“My role is to facilitate access to data from various business units,” Gawor said. “We see it as an opportunity for these various business units to tell their story.”

He also works with non-profit groups and various community groups to collect data.

Each topic expands on the raw data by providing context. In the case of goals, there will be a section of the article answering what the City of Calgary is doing to address the issue, which range from waste management to property assessment.

The Public Sector Digest ranked Calgary the fourth best “Open City” in 2016 while Edmonton placed first. 

“It speaks to the readiness, implementation and the impact of our data,” Gawor said. “We’re continuously improving, but we are putting out valuable datasets.”

While the Open Data Portal isn’t focused on machine learning algorithms, it provides raw and free data for the public and data scientists to utilize. Gawor explained that the intellectual property resulting from any applications produced belong to the user rather than the City of Calgary.

“Without data, you can’t make predictive models, you can’t have business intelligence, so we’re sort of the enablers by providing access to those datasets that people can start looking at,” he said.

Currently, citizens can leave suggestions for datasets they would like to see, but Gawor said it would be exciting if they could personally contribute data, like locations of potholes, in the future.

He believes the visualization of data will be an important tool for not only the general public, but decision makers as well.

“I’d love to see more and more data being used to drive decision making in council and even just for citizens for their daily needs,” he said.

In terms of AI and data science in general, Gawor has high hopes for both the city and province.

“It’s something to be proud of that in Alberta we have this level of technological interest and support,” he said.

The global innovator: How Watson helps industries worldwide

Rob Burton, cognitive solutions specialist with IBM, trained Watson for the natural resource industry. Photo by Paul McAleer

IBM is a global leader in AI innovation and is the mastermind behind IBM Watson, which was introduced back in 2011.

Watson is a cognitive system based on four principles relating to the consumption of data:

  • Understanding
  • Interacting
  • Reasoning
  • Learning

Experts Define AI
“Artificial Intelligence to me is artificial. It implies there’s a fakeness to the intelligence. We’ve classified AI as augmented intelligence. Humans are intelligent, we’re cognitive, we’re the ones that do all the thinking from a large perspective.” — Rob Burton, IBM

In Alberta, Watson is being used to predict schizophrenia in Edmonton and to assist start-ups and entrepreneurs through the District Venture and IBM Innovation Space in Calgary, as well as working with businesses like Goldcorp Inc.

Watson first appeared on the American television game show Jeopardy! By sifting through information already stored in its memory the computer defeated two Jeopardy! champions without being connected to the Internet.

The idea for Watson stemmed from Deep Blue — a chess-playing computer that eventually defeated world champion Garry Kasparov in 1997 — and the need for a new challenge to put computing technology to the test.

Fast-forward to 2017 and Watson is now being used to help a variety of industries, including healthcare, financial services, law, retail, education and the natural resource sector. It’s trained in eight languages other than English and IBM predicts Watson will interact with approximately one billion consumers by the end of the year through the businesses it helps.

Along with Apple, Facebook, Microsoft, Google and Amazon, IBM is a founding member of the Partnership on AI, an initiative for matters such as ethics, bias, transparency, economics and privacy.

Watson in Alberta

Rob Burton is a cognitive solutions specialist at IBM’s Natural Resources Solution Centre in Calgary, and he’s been working closely with Watson since he joined the team two years ago.

As a Star Trek fan, Burton was fascinated with AI in the 80’s and wanted to build future cities in space. He never expected to work with AI on the same level as Watson, but he sees a lot of similarities between machine learning and developing a space city.

“If I’m building a space city, you have to understand every single process that’s involved with it. It became a very complex puzzle,” he said. “If you don’t have a sequential systematic approach to executing the code, you’re not going to have any responses.”

With a background in complex process systems, Burton was integral when it came to providing Watson with the necessary framework to analyze data relating to the natural resource sector.

“One of Watson’s brilliant services that we’ve developed right up front is called the discovery service. It allows Watson to read unstructured documentation, which is the first for any type of system of its kind,” he said. “It can read my [Microsoft] Word documents, it can read my PDF files, it can read websites, so it can read all the unstructured things just the way a human can read it and understand the context, the nuances, the idioms — all of those details about what that unstructured information is.”

Is AI Dangerous?
“Everything we’ve done throughout history has pros and cons. What it comes down to is the individual experience. If I’m going to use it for bad, then you could do that, but it’s up to the individual. It’s the human that’s basically trying to make decisions from the augmentation of information.” — Rob Burton, IBM

Perhaps most importantly, Watson is capable of understanding data from a multitude of perspectives. For example, an engineer and geologist may have different views about data and misunderstandings about each other’s roles.

“With our predictive and cognitive insights that we add through Watson, we can find cross-informational insights from both sides,” Burton said. “We’ll look at the same set of documents and it might come back with 5,000 [insights] for the geologist and 10,000 for the engineer.”

From there, Watson will come up with ranked recommendations for the client to assist with operational decisions.In the case of Goldcorp, Watson analyzed a vast supply of geological data to help the company improve mining efficiency.

Watson also worked with Woodside Petroleum and the company had high praise for its AI team member:

“Woodside said, ‘we taught Watson to be an engineer, but Watson taught us to think like a thousand engineers,’” Burton said.

IBM Watson and related facts
Watson can communicate in nine languages, including Japanese, Spanish and Italian.

Watson solutions are applied in more than 45 countries, in 20 different industries.
The International Data 

Corporation (IDC) predicts 75 per cent of all consumers will interact with cognitive computing by 2018 and in 2020 spending in the area will surpass $40-billion.Even though Watson is a valuable tool for nearly any job, it takes a tremendous amount of resources to train it.

“An AI can’t work independently without a human,” Burton said. “It needs direction, it needs instructions, it’s always learning. It’s kind of like the forever child — it will never ever stop learning.”

Depending on the task, Burton said Watson needs knowledgeable teachers in order for it to provide effective feedback.

“When I need to give advice and an engineer a recommendation, it has to have the same caliber of education the engineer has,” he said.

As for the future of Watson, Burton would like to see its uses continue to expand at both a business and personal level.

“If I could just have a personal assistant, which we’re working on, that I can put on my system and just say, ‘I’m interested in these top news stories today. Go out and find them for me,’” he said. “I really see having an augmented intelligence or having something that can do the heavy lifting and research for you as a value for any client at any level.”

This feature story can also be found in the Calgary Journal print edition that hit shelves on Nov. 9, 2017. Looking for a physical copy? Please check these locations.

pmcaleer@cjournal.ca

Editor: Tyler Ryan | tryan@cjournal.ca