Canada is open for AI business – some fear too open

The world’s tech powers are sending giant sums of money spinning into Canada, but while many see this as a sign of success, others are worried about researchers and intellectual property being swallowed wholesale.

The country is in the midst of an artificial intelligence (AI) boom, with Google, Microsoft, Facebook, Huawei and other global heavyweights spending millions or even hundreds of millions of dollars on research hubs in Quebec, Ontario and Alberta.

Canadian doors are open – some fear too open.

Jim Hinton, an IP lawyer and founder of the Own Innovation consultancy, reckons that more than half of all AI patents in Canada end up being owned by foreign companies.

“We’re selling our lunch. What we need to be doing is getting money out of our ideas ourselves, instead of seeing foreign talent scoop it all up,” said Hinton. “Otherwise we’ll never have a Canadian champion.”

The country is home to hundreds of fledgling AI companies, including much-talked-about start-ups like Element AI and Deep Genomics, but they remain relatively small. “They don’t have a strong market position yet,” Hinton says.

Deep learning pioneers such as Yoshua Bengio and Geoffrey Hinton (no relation to Jim) have nurtured top-notch talent in AI in Canada for years, back when AI was an emerging field.

The country today ranks in the top five worldwide for the number of high-impact AI researchers and for AI-related job openings, according to the Global AI Talent Report 2019.

But despite Canadian inheriting this brilliant AI lead from the country’s AI “godfathers”, big foreign players have an unassailable advantage over homegrown efforts, Hinton said.

“It’s not an easy go for the average company to make a business out of AI. It’s about the datasets you have. If you have great access to data, you can outperform anybody, any day of the week,” said Hinton.

He would like to see the government introduce a mechanism such as exists in France, setting strict oversight of foreign bids for companies in key industries.

“The Canadian government confuses a job strategy with an innovation strategy,” he said. “Industry has moved on from the 1970s when building ships would get you all these spillovers, with guys across the shipyard making rubber grommets and so on. The spillovers from AI aren’t so many; they aren’t so tangible.”

Hinton points to the government’s $49 million subsidy to MasterCard to set up a new technology centre in Vancouver. “Great for jobs but it’s handing an economic advantage to a foreign company,” he said.

Toronto is booming

Many others say Canada’s willingness to experiment and put government, academic and commercial resources behind AI is having a huge economic and skills payoff.

“We’re here to help superstar researchers but our main goal is to strengthen the region,” said Garth Gibson, CEO of the Vector Institute, Toronto’s public-private artificial intelligence research institute.

Toronto has never seen momentum like it’s experiencing right now, Gibson says, with the dollars pouring in.

“You have to have a strong workforce, a compelling place to work, a good tax law and a sense of competitiveness among your people. We have all of that,” Gibson said. In 2017, according to government statistics, the city added 30,000 tech jobs; more than the San Francisco Bay Area, Seattle and Washington, DC combined.

Vector is booming too: some 50 companies, including Google, Uber and Thales and big Canadian companies like the Royal Bank of Canada, Scotiabank and Air Canada, pay a high premium to work closely with academic talent – C$300,000 dollars on average for a five-year membership.

“The companies invest a fair amount of direct cash, but also the time of their personnel as well,” Gibson said.

Paying big dollars for AI academics is common, he said. “Young professors who are not yet tenured – they could be looking at a salary of more than half a million dollars,” Gibson said.

Joelle Pineau, managing director, Facebook AI Research Montreal, and associate professor, School of Computer Science, McGill University, says this is benefitting the field of AI as a whole. “[We are] deeply integrated with Canada’s thriving AI ecosystem and our investments allow us to foster and retain top Canadian talent, elevate the training environment, and advance the overall progress of AI research,”

A lot of the money being invested in Vector is going back into computer resources, with the research body investing C$4 million in servers alone in the last few years, says Gibson. Geoff Hinton serves as the institute’s chief scientific adviser.

Gibson says that start-ups will always struggle to compete for staff against big tech companies. “When you’re starting a firm in a small market, it certainly feels like big companies have undue power. I started a tech company in Pittsburgh years ago and we were competing for hiring with Silicon Valley so of course it was hard. You’re going to be hurt by winners,” he said.

“But in all this, the most important thing for me is seeing the R&D done in my region. I would love the profit to stay as well of course,” Gibson said. 

Another big AI fixture in Toronto is the Creative Destruction Lab, a mentoring programme started in 2012. Bogdan Knezevic, who leads the lab’s AI work, said the motive for founding it was to transmit entrepreneurial skills.

“It wasn’t the lack of capital, or the lack of good ideas in the city. What was missing is what has existed for so long in the Valley – the handover of knowledge from one person to the next,” he said. The nine-month programme pairs company founders with experienced entrepreneurs and investors to set business objectives.

“We’re able to attract people from the Valley; we have people flying in from Hong Kong. It’s a telling sign,” Knezevic said.

Toronto is not the only Canadian city seeing significant AI action. Microsoft and Samsung are growing research hubs in Montreal. DeepMind, possibly the most famous AI company in the world, whose system was famously the first to defeat a master of the complex board game Go, has an office in Edmonton.

Big advantage

Gilles Savard is general director of the Institute for Data Valorisation, a lab of over 1,000 scientists from HEC Montréal, Polytechnique Montréal and the University of Montréal, doing research and technology transfer, covering data analytics and machine learning.  

“We connect companies to knowledge and talent,” said Savard, a specialist in mathematical programming. “We’ve got a quarter billion dollar budget to spend over six years – a lot of money.”

The might of the American tech companies is a problem for everyone, Savard says.

“We should be careful about the GAFA [Google, Apple, Amazon and Facebook] companies. I am concerned about their size,” he said. The four companies were contacted for comment. He compares their hegemony to the oil industry in 1911, when the US Supreme Court ruled that Standard Oil was too big and needed to be broken up. “Well, here we’re talking about the oil of this century: data,” Savard said.

The American tech companies face growing concerns around the world over the large amounts of personal data they are storing and sharing with outside businesses. “The big four have more power than many, many governments. They should have to pay taxes in all the different countries they operate in. For me, that’s obvious,” said Savard.

“They’re getting too big and we need to have a good reflection on that,” he said. “Though we realise the US President is not the type of guy to put a brake on the GAFA’s development.”

Foreign assets

Others say Canada’s drive to develop an AI industry owes something to the Trump administration.

Canadian researchers say they receive a flow of inquiries from their counterparts in the US, concerned about the White House’s stance on immigration and other policies.

For disillusioned scientists in the UK, political unrest around Brexit has also helped make Canada appear a more attractive destination.

“We have recognised a wonderful opportunity to bring researchers from UK and US,” said Elissa Strome, executive director at the Canadian Institute for Advanced Research, which runs the country’s $125 million national AI strategy.

The institute has helped headhunt 80 research chairs with funding of around $1 million over five years. The chairs take up positions in Canada’s three national AI Institutes – Amii  (Alberta Machine Intelligence Institute) in Edmonton, Mila in Montreal and the Vector Institute in Toronto.

The funding effort is about recruitment – the programme has also targeted researchers from countries including China and India – but also retention of the country’s big research names, Strome said.

“The concern was that we were training very talented graduates and then losing them. Major tech firms were recruiting them away from us. So we wanted to keep what we had here, build on it, and reverse the flow outwards,” Strome said.

Vancouver’s traffic edge

One of the recruited chairs is Angel Chang, a member of Amii, and an assistant professor at Simon Fraser University in Vancouver.

Chang, who was born in China but has lived most of her life in the US, works in natural language processing and 3D visualisations. “I want to build robots that understand how to go to the kitchen and get me a banana,” she said.

One of her main reasons for choosing Canada was to escape the notorious traffic in Silicon Valley. “In the mornings, I would drive to Facebook first to drop off my husband, and then go on to my start-up,” said Chang. “It would generally take one hour for me to get to the next motorway exit, a journey that should really take 10 minutes. At some point I realised I just couldn’t spend one or two hours in traffic every day,” she said.

“The Bay Area weather is the best in the world but there are some drawbacks. You get into a bubble where people think very similarly. Also, it’s becoming overcrowded and the housing is ridiculously expensive,” she said.

Though initially wary of Canada’s cold winters, she chose Vancouver, where the AI scene is smaller than in Montreal and Toronto. “It’s much more calming,” Chang said. “And there’s public transportation.”

She has noticed a difference in Canadians' outlook. “In the Bay Area, people have the start-up mentality. Failing is good. There is still the stigma of taking risks if you go somewhere else,” she said. “The culture here is less entrepreneurial, but I do like being in the freedom of academia.”

Canada’s early funding for path-breaking opened the door to remarkable improvements in AI technology, said Chang. The government is keeping up its interest, with the creation of five innovation superclusters, one of which aims to harness AI and robotic technologies to transform supply chains. Officials hope it will add more than C$16 billion to the economy over 10 years.

“The government has shown a willing to fund the kind of research that people used to worry was a dead end,” said Chang. “By doing this, they were telling everyone that they are willing to fund things that are very long term.”

Step up

There’s nothing that makes Canada particularly special for AI, argues Martha White, an assistant professor in the University of Alberta’s Department of Computing Science, who is also an AI chair.

“I can’t help but feel it’s not different than anywhere else, but I guess the research community is smaller, and that sometimes makes it more cohesive,” she said. “And then I suppose there’s the godfathers, who were doing AI back in the 80s and 90s when it wasn’t cool. They’ve had a big political impact.”

The benefit of the AI chair network, she says, is that it can lobby the government, as it did last year on behalf of the AI experts from Africa and South America who were denied visas to attend an industry conference in Vancouver.

Overall, she thinks the government is doing a good job though. “When I look at the world and see what should be funded, I don’t think we deserve more,” she said.

White also doesn’t view the presence of giant foreign companies in Canada as a threat for the development of homegrown ideas. “The reasons Huawei and others are coming forward is in part because our businesses are not stepping up,” White said. “There isn’t the same level of commitment from businesses here on AI – they’re a little conservative, and I would never say that about American companies, which jump on the hottest new thing.”

Businesses “should be looking for expertise, within universities, and be willing to use AI tech to transform themselves, even if feels scary,” White said.

The tech giants are paying their fair share, she says. “They come and fund a lot of basic research; they fund chairs; they are investing in infrastructure. They’re willing to be a bit speculative,” White said. 

Canada has around 650 AI start-ups. The question is whether these companies will evolve into vibrant independent companies, or just disappear into a company like Apple or Google.

“If you’re going to get to be a unicorn, you need hundreds of millions of dollars and there’s not as much VC here as there is in San Francisco,” Gibson said.

“But nothing is pre-ordained. I remember Google coming in – they were late to the search space, look at them now. Amazon nearly went away many times. Facebook started out as a college prank.”

The responsible checklist

One niche for Canada could be in setting a global standard for AI ethics development.

Researchers like Catherine Régis are trying to help Montreal lead the world in ensuring AI is developed responsibly.

“AI is slowly moving from soft law to legislation,” said Régis, professor at the University of Montreal’s faculty of law and co-author of the Montreal Declaration, a series of principles seeking to guide the evolution of AI in the city.

The declaration is meant to be a collaborative project between computer scientists, researchers, lawyers and others. Its principles are broken down into seven themes: well-being, autonomy, justice, privacy, knowledge, democracy and responsibility.

“We had very tough discussions on the feasibility of the principles but I think it allows companies to anticipate the future. It’s an ethical compact to give you everyone some food for thought. It should be a checklist for researchers and for companies,” she said.

Checklists like this one are evidently not influencing AI use everywhere. In China, for instance, AI technologies are in the service of developing what critics fear is a high-tech authoritarian future.

Chang says the path taken by the country where she was born leaves her with “conflicting feelings”. “On the one hand, it’s great to see China emerging as this very competitive economy; I feel proud about this. But there are some concerns, when we can see the government is closing in and restricting its citizens. AI is even being used to shame people who wear pyjamas on the street. More seriously, it could potentially land you in jail,” she said.

Canada’s set of values are such that it won’t go down a route like China, Savard argues. “Could we ever put a score on people in Canada? Rank our citizens using AI? Impossible to imagine it,” he said.

Some of the work in Savard’s lab is to figure out a way of certifying AI systems where the underlying logic of the technology is not easily legible for humans.

Today, an airplane runs on lines of code that are relatively easy to understand, test and validate, but increasingly intelligent machines however pose a new reality with their “black box” decision-making, says Savard. Humans are turning themselves over to the unknown.

“We don’t know which new data will come into these algorithms. And we don’t know what the outputs on the other end will look like,” he said.

In the future pilots flying planes that run on some kind of AI software, will need to have “an understanding of why an algorithm suggests something,” said Savard. “You need built-in explain-ability and transparency. That’s going to be really hard.” Similarly, doctors will want to know why an AI-powered device tells them to take a particular course of action.

There is a new mindfulness around the power of technology, Savard says. “We’re not talking about hindering AI, but there has to be some rules,” he said.

Some assembly required

Savard counts himself among the AI optimists. “We’re moving towards a more intelligent use of data. We can achieve a longer and better quality of life through personalised health. Digital intelligence can make transport and energy much more sustainable,” he said. “I may be naïve but I think AI can help us,” he said. 

It will take some time still to get here. “We are quite a while away from being able to command robots, for example,” said Chang. She explains the limitations of machine speech recognition. “The difficulty is that my robots have many more rules to learn and follow than self-driving cars, for example,” she said. “If you want to tell your indoor robot to clear the table and put away the dishes, that’s a fairly simple statement, but for the robot to execute, it needs to know a lot of information,” she said.

For example, do the dishes need to go in the dishwasher? Can the robot figure out how to open the dishwasher? Can it place dishes in neatly without smashing them? “Then it needs to close the door and push the ‘on’ button,” Chang said.  

The world has got as far as voice command systems – Apple’s Siri, Amazon’s Alexa and Google’s Assistant – but even they have some failings.

“My Siri didn’t know I had moved to Canada, so when I asked it ‘what’s the weather like today?’, I got updates for California,” said Chang.