With U.S. giants such as Netflix Inc. and Amazon.com Inc. threatening his business, Doug Murphy knew he had to play their game to compete. So last year, the chief executive officer of Corus Entertainment Inc. asked Integrate.ai Inc., a Toronto artificial-intelligence startup, if it could help the Canadian broadcaster use its trove of audience data to fight back.
Corus and Integrate.ai soon settled on a project: deploying AI to get viewers to watch more shows on its channels. Corus executives typically made gut decisions about when to run cross-promotional commercials, but they tended to be obvious – advertising one gardening show during another on HGTV, for instance. What if Integrate.ai’s algorithms could root out unforeseen correlations? What if, for instance, it could unearth the fact that History Channel regulars would also be more likely to turn to the Food Network to learn how to make soufflé?
The code went to work, and when Corus took Integrate.ai’s first suggestions, they were 50 per cent more effective than past efforts at generating viewership for shows such as Siren on ABC Spark, which was promoted during commercial breaks of The Incredible Dr. Pol on Nat Geo Wild. “There’s clearly something there,” Mr. Murphy said. “If I can move the needle on the efficacy of cross-promotions, that creates real value for me.” Now, he said, “more light bulbs are going off” inside Corus about how to solve other business challenges using artificial intelligence tools.
Corporate Canada is starting to wake up to AI. Name any sector and you’ll find at least one established player that has experimented with it – not the stuff of Terminator or 2001: A Space Odyssey, machines with human-level consciousness and perception, but software in narrower areas of machine learning.
Artificial-intelligence technology has advanced enough that algorithms can perform as well or better than humans at recognizing speech and images – and outperform us at solving certain problems or predicting outcomes. “I think [corporate executives] are starting to see a disadvantage to their business if they don’t have an active, applied machine-learning or AI project that is delivering results,” said Integrate.ai CEO Steve Irvine.
Many observers believe AI will eventually become as widely used across the corporate world as the personal computer in the 1980s and e-mail in the 1990s. IDC, a market research firm, estimates worldwide spending on “cognitive and artificial intelligence systems” will rise to US$58-billion in 2021 from US$8-billion in 2016. Accenture has forecast that businesses that apply AI could increase their profits 38 per cent on average by 2035 and collectively raise their value by trillions of dollars. Queen’s University is even launching North America’s first master of management in AI program this fall.
The controversy surrounding the misuse of Facebook personal data, last month’s fatal collision involving a self-driving Uber car and impending data privacy regulations in the European Union may temper the proliferation of AI and raise a lot of questions. But they won’t stop it.
That’s good news for Canada in one respect. Canadian AI experts such as Geoff Hinton led the way in developing the science of machines that can learn. But it was Silicon Valley that took the lead in commercializing it for the tech sector and hiring away many top Canadian-trained minds.
But now, Canadian entrepreneurs are leading a burgeoning homegrown industry that aims to take Canada’s AI reputation out of the lab and into the marketplace, selling AI for everyone else in the corporate world.
Canadian AI startups are getting audiences with executives at large firms eager to try out the technology. “Over the last six months we’ve gotten a lot of calls from larger and larger companies,” said Aran Hamilton, CEO of Toronto-based Vantage Analytics, a six-person firm that has helped global packaged-goods companies such as Unilever NV with strategies to increase e-commerce sales, order sizes and repeat purchase rates. “They’ll just say, ’We’re looking for someone to help us. How fast can you start?’”
Venture capitalists have taken note, staking huge sums on Canadian AI startups including Integrate.ai, Rubikloud Technologies of Toronto and Montreal’s Element AI, co-founded by Montreal professor Yoshua Bengio, which raised US$102-million in 2017 from a who’s who of global investors. Microsoft, Google and Toronto-Dominion Bank have shelled out for startups to acquire their AI talent.
Competition to hire those scarce experts and for the emerging class of AI vendors is already intense. Many tech giants provide AI software of their own, including Alphabet Inc.’s Google, International Business Machines Corp., Salesforce.com Inc. and General Electric Co. Dozens, if not hundreds, of other AI startups in hot spots such as Silicon Valley, New York and China have raised billions of dollars in funding as well.
Meanwhile, many startups are beginning to achieve results for a range of corporate customers, some of whom have started to make a point of highlighting that work to investors.
Major corporations from Telus Corp. to Bank of Montreal have adopted automated customer service chatbots from Toronto’s Ada Support Inc. and Vancouver-based Finn.ai, to help give faster answers to customers who ask questions online. Other banks, retailers and consumer packaged goods giants have worked with Integrate.ai, Rubikloud and Vantage to send customized promotions to prospects. Manufacturing plants in Canada, the United States and Europe are using AI-enabled tools from Toronto’s Canvass Analytics and Ottawa’s Raven Telemetry to predict when machines will break down, fix quality-control problems or optimize energy use. Mindbridge Analytics Inc. in Ottawa sells tools that flag improper financial transactions for human auditors.
Several Canadian startups now talk confidently about becoming the country’s first AI-focused business to crack the nine-figure revenue mark. “We know we have the right product, the right market problem and the opportunity to be a $100-million company in the next three to four years,” said Kerry Liu, CEO of Rubikloud, a five-year-old firm that raised US$37-million in venture funding early this year, led by Intel Capital.
Despite the bravado, though, Canada’s AI startups must overcome a big challenge: convincing companies to take a chance on something that many still don’t fully understand.
Canvass CEO Humera Malik said some companies have approached her saying, “’We want to do an AI project.’ We ask, ‘What do you want to do?’ And they say, ‘Well, we don’t know.’ And you have to handhold them through the process.”
A recent global study by McKinsey & Co.’s research institute found that AI adoption outside the tech sector “is at an early, often experimental stage.” Another upcoming McKinsey survey of 120 senior Canadian executives found that 89 per cent believe AI will create major positive change within five years – but little more than a third have changed their longer-term strategies to incorporate it. Despite living in one of the leading countries for AI development and research, many Canadian executives are confused about how to effectively use the technology, and there are “notable gaps” in their understanding of AI’s potential and impact on their businesses, the study said.
’The beginning of a long transition’
Widespread adoption of AI will bring huge changes in the ways companies operate their businesses and make decisions. It will be an era in which the cost of making predictions will tumble, as machines do that work efficiently, while the value of human judgment increases, say University of Toronto business professors Ajay Agrawal, Joshua Gans and Avi Goldfarb in their new book, Prediction Machines: The Simple Economics of Artificial Intelligence.
“No wonder corporate directors, CEOs, vice-presidents, managers, team leaders, entrepreneurs, investors, coaches, and policy makers are anxiously racing to learn about AI: they all realize it is about to fundamentally change their businesses,” they write. “There is often no single right answer to the question of which is the best AI strategy or the best set of AI tools, because AIs involve trade-offs: more speed, less accuracy; more autonomy, less control; more data, less privacy.”
Many in the corporate world appreciate the rationale for using AI: They believe they have no choice if they want to avoid being left behind. “We’re seeing the beginning of a long transition that all large organizations that want to remain relevant have to make over the next 10 years,” said Jean-Sebastien Cournoyer, co-founder of Real Ventures, one of Canada’s leading venture capital investors in AI.
The Silicon Valley juggernauts that were the first to adopt AI – such as Amazon and Apple Inc. – have built massive advantages that have transformed entire sectors. Their AI-based technology isn’t just powering searches, recommending content and running digital home speaker devices; it is helping them make commercial use of the explosion of data generated by smartphones and connected sensors.
Amazon’s recommendation engine continues to get sharper, to the detriment of already-threatened traditional retailers; Apple is improving its autocorrect feature; and Google has begun suggesting automated e-mail replies to Gmail users – all thanks to AI. Netflix has experimented with generating personalized movie trailers based on the preferences of individual subscribers.
As they collect more data, the internet giants know more about what consumers want –and when and where they want it – disrupting traditional customer relationships, says David McKay, the CEO of Royal Bank of Canada.
“I see [customer] intent at the end stage, but as they go through that thought process, I’m not always privy to that. I used to be,” Mr. McKay said. “People came in and talked to their banker about the financial implications about what they were doing before they made decisions. Now [they] go online and plan those decisions … and others are intercepting and reading those signals. They may use that information to offer that service themselves.”
“A lot of large companies feel they have been under attack by new technology players,” said Janet Bannister, a general partner with Real Ventures. “And [the established companies] view AI as their opportunity [to] have the upper hand because this time they have all this data.”
Getting to that data, however, is not easy. Many companies are only starting to appreciate how much data is trapped inside their systems and machines. Ms. Malik of Canvass said some manufacturing plants produce tens of millions of data points per minute – data that, once analyzed by intelligent algorithms, could lead to a better way of making things or significant savings. (In one case, she said, her company has saved an auto parts manufacturer millions of dollars a month by using AI to pinpoint the cause of a production flaw.)
That data must first be identified, extracted, collected and presented in a way the algorithms can make sense of it, however. For some companies, that’s an undertaking that could take months.
A handful of large companies here started early. RBC opened an AI lab two years ago in Toronto, headed by chief scientist Foteini Agrafioti, an inventor and entrepreneur who founded biometric wristband maker Nymi after earning her PhD from the University of Toronto. The RBC operation, called Borealis AI, now has dozens of people at labs across the country conducting both fundamental and applied AI research, including developing algorithms to prevent fraud and predict how events such as severe weather could affect markets. In those two years, RBC has collected more data than in the previous 20, Mr. McKay said.
But with AI talent in short supply, many other big corporations will be looking to AI startups to do the job.
The Canadian AI startups and their customers say their early engagements have generated promising results. Shannin Hudson, the plant manager of a dental instrument factory in Morrisburg, Ont., that is owned by U.S. giant Danaher Corp., said her operation increased “up-time,” or the proportion of time that the machines are operating, to 92 per cent from 78 – saving hundreds of thousands of dollars annually. The plant used Raven’s system to gather data from 150 machines that had never tracked before, pinpointing when and why they weren’t performing. Bank of Nova Scotia’s Chilean credit card business, Cencosud, increased revenues in one marketing campaign by 40 per cent after Integrate.ai applied machine learning to predict how customers would respond. “Being really efficient on our customers’ behalf is just really good business. We believe that AI can help us to do that,” said the bank’s chief digital officer, Shawn Rose.
Canada’s AI startups must now demonstrate that their success is not limited to a few test cases. “The question has shifted from how enterprises can leverage AI technology to how they can scale it across their businesses,” said Stephen Piron, co-founder of Deep Learning, a Toronto AI startup that has also worked with Scotiabank. “A C-level exec at one of our clients challenged us by saying, ’It’s fantastic your AIs are creating $10-million in value, but how do we get this to $100-million?”
If they can do that, the opportunity is huge, Mr. Murphy said. “Their addressable market is traditional industries that need to reinvent themselves in the digital age, and that’s basically everybody.”