TORONTO – The infrastructure Canada has created for artificial intelligence (AI) research is at the top of its class, but needs maintence to remain that way, an executive who left both New York City and Silicon Valley to work at Toronto’s Integrate.ai says.
In particular, during a Jan. 31 panel titled “Beyond the Hype Cycle: Deploying AI in the Real World” at the 2018 Cantech Investment Conference, Integrate.ai vice president of product and strategy Kathryn Hume praised the Canadian government’s prescient support for the Canadian Institute For Advanced Research (CIFAR)’s AI research in the 1980s and 1990s, which helped create an ecosystem that attracted such luminaries as University of Toronto researcher Geoffrey Hinton and McGill University’s Yoshua Bengio, and the University of Alberta’s Richard Sutton, who in turn have served as a magnet for the world’s foremost AI researchers and partners – including herself.
“I’ll often give talks and use this slide that has the Italian flag with Marlon Brando in the middle, that then replace his face with Geoffery Hinton’s, because in the 80s, CIFAR was known as the ‘Canadian Mafia’ – they had created the place where [machine learning] researchers went to die so they could work with the likes of Geoffery Hinton, Yoshua Bengio, and Rich Sutton – the three godfathers of the movement,” she says.
Unfortunately, Hume hears few discussions about how to maintain that ecosystem, which siphons too many of its best and brightest to California after graduation.
“There needs to be more dialogue,” she says. “There’s amazing work going on right now in the theoretical research realm, and you’ve got companies like Uber and Samsung and Facebook and DeepMind – all of these large global companies – setting up research labs here. And that’s great! Fantastic! But as a startup person it’s limiting, and I believe we need to have a stronger dialogue around other ways it could work and what should come next.”
It’s not that citizens of the global AI ecosystem have no reason to choose Canada over Silicon Valley – after all, Hume is one of them. Prior to arriving at Integrate.ai, developers of an AI-powered customer service platform for B2C enterprises, she served as the director of marketing at New York City’s Fast Forward Labs (later acquired by Cloudera) and as a marketing associate at San Francisco startup Vantage Local Inc.
“When I told friends of mine in the machine learning community in New York and San Francisco about my decision to move to Toronto they were like, ‘of course,’ because the people who go to the ICML (the International Machine Learning Conference) and NIPS (Neural Information Processing Systems) conferences get that coming to Canada connects you to one of the world’s best machine learning networks these days,” she says.
Integrate.ai founder and CEO Steve Irvine made the choice between Toronto and Silicon Valley as well, having run Facebook and then Instagram’s marketing partner programs from its Menlo Park headquarers between 2013 and 2017.
Maintaining the ecosystem that pulled her and Irvine north will require collaboration between multiple sectors, Hume says, each of which brings their own advantages.
The education pipeline, for example, needs to be better educated about the opportunities provided by startups, which Canadian venture capital needs to be quicker to support.
“The social capital of academia is different from that of government, which is different from that of enterprises, which is different from that of startups,” she says. “The ideal world would be where all are somehow able to communicate and collaborate.”
“There’s a bit of that dialogue, but not enough,” Hume continues, acknowledging that it’s easier said than done.
However, she believes it’s possible – and more so in Canada than in the U.S.
“I don’t see the same sort of openness for thinking about policy from a startup perspective in the U.S. that I see going on here,” Hume says. “But we’ve still got a long way to go.”