CIFAR is keeping Canada at the forefront of AI
The 19th-century facade of the Power Corporation of Canada headquarters in Montreal belied the cutting-edge technology being discussed inside on a Tuesday evening this past April. Some of Canada’s most prominent business leaders were hearing from a selection of the world’s foremost experts on artificial intelligence (AI), including CIFAR Senior Fellows Yoshua Bengio and Joëlle Pineau, and celebrating the tremendous success of the Québec AI ecosystem.
CIFAR Senior Fellow Yoshua Bengio at CIFAR's Montreal AI leadership event. (Photo credit: Pat Pinotti, Les Productions Studio 55 Inc.)
The success in Montreal is a microcosm of what’s going on nationally, thanks in part to the Pan-Canadian Artificial Intelligence Strategy and CIFAR’s leadership in AI. It’s an example of how a strong partnership among research, government and industry can retain and attract the world’s top talent and keep Canada at the leading edge in a field that will significantly shape the way we live for years to come.
CIFAR Senior Fellow Joelle Pineau at CIFAR's Montreal AI leadership event (Photo credit: Pat Pinotti, Les Productions Studio 55 Inc.).
Canadians and Canadian-funded research have long been at the forefront of developments in AI. CIFAR recruited Geoffrey Hinton, now a CIFAR Distinguished Fellow and chief scientific advisor at the Vector Institute in Toronto, to Canada in the 1980s. At the time, most AI researchers saw neural networks as a dead end. But Hinton started what is now CIFAR’s Learning in Machines & Brains program, and the foundational work he and others did within the program helped spark the deep learning technology we know today. Now, deep learning fuels everything from smart cars, to personalized medicine and diagnostics, to science itself.
So when the federal government announced its $125 million investment in the Pan-Canadian AI Strategy in March 2017, it looked to CIFAR for help.
"CIFAR’s long track record of supporting pioneering AI research and success in bringing together research leaders from across disciplines and around the world to tackle tough problems made us the obvious choice for delivering on the goals of our national AI research strategy,” says Elissa Strome, executive director, Pan-Canadian AI Strategy at CIFAR.
The strategy aims to cement Canada’s international leadership in AI research and training through four primary areas. The first is to establish three centres of excellence in AI research across the country: Vector Institute in Toronto, MILA in Montreal and Amii in Edmonton. The institutes’ leadership includes current and past members of CIFAR’s Learning in Machines & Brains program, from Rich Zemel in Toronto, to Bengio in Montreal, to CIFAR Associate Fellow Rich Sutton in Edmonton.
The second, and hallmark activity, the Canada CIFAR Chairs in AI program, is dedicated to recruiting and retaining leading AI researchers to Canada as well as fostering the next generation of talented AI scientists. The program will see 50 Canada CIFAR Chairs in AI appointed across the three institutes. CIFAR is working closely with the AI institutes as they identify possible candidates, and in February appointed seven members to the Pan-Canadian AI Strategy’s International Scientific Advisory Committee. This committee, comprised of internationally distinguished individuals from leading universities and companies from the US, the UK, France and the Netherlands, will review nominations for the CCAI Chairs program and provide strategic advice and guidance on the overall strategy.
Already, support from federal funding, and the significant provincial contribution ($50 million in Ontario and $100 million in Québec) and private sector investment is drawing AI leaders to Canada. Carnegie Mellon professor Garth Gibson has been appointed CEO of Vector. The newly established Microsoft Research Montreal recruited Geoff Gordon, also from Carnegie Mellon, and CIFAR Senior Fellow Alán Aspuru-Guzik left Harvard to join Vector and the University of Toronto as a Canada 150 Research Chair.
Further investments have focused on expanding the pool of talented individuals in Canada who have skills and expertise in AI. Last October, the Ontario government increased its commitment to AI training, investing an additional $30 million in Vector over five years to work with universities to build a strong cohort of AI master's students.
The AI & Society Program lets us tackle the questions that are important, but may not necessarily be getting attention right now...
“The Province of Ontario saw an important opportunity to respond to demand from students and employers alike for expanded training opportunities in AI. The goal is that by the end of five years, Ontario will be graduating 1,000 master's students per year in AI-related fields. That kind of density of talent can lead to significant positive economic impact,” says Strome.
The third component of the strategy is the creation of a national AI program of activities to encourage greater coast-to-coast collaboration in research. These activities include workshops, a December meeting of the CIFAR Chairs in AI in Montreal and a series of hands-on summer schools dedicated to helping young people, from high school to graduate school and beyond, develop skills and expertise in AI that will be beneficial to their careers.
Two summer schools worth highlighting are the upcoming annual CIFAR Deep Learning/Reinforcement Learning Summer School, hosted by the Vector Institute and University of Toronto this July for 250 graduate and post-graduate trainees; and McGill University, MILA and the OSMO Foundation’s AI for Social Good Summer Lab, which provides women with machine learning skills and practical experience to increase their career opportunities.
The national AI program’s workshops, Aix (or AI to the power of X), will also provide opportunities for researchers to get together with machine learning and AI experts to think about how to apply AI approaches to their particular domain. One workshop this January with the CIHR's Institute of Population and Public Health brought together public health researchers and AI researchers. A second workshop, AI for Neuroscience, is coming this fall.
“The Tri-Council Agencies just launched a call for proposals for AI, Health and Society projects so we’re already seeing outcomes as a result of the workshops that CIFAR helped to organize and facilitate,” says Strome.
The fourth and final component of the strategy is perhaps the one of most interest to a media-consuming general public already intrigued by how AI will impact our daily lives. CIFAR’s AI & Society Program will support research and policy outreach on the ethical, legal, economic and policy opportunities and issues AI presents for society by convening working groups and publishing the results in non-specialist publications that are accessible to a broad audience.
"The AI & Society program lets us tackle the questions that are important, but may not necessarily be getting attention right now. It also gives us the chance to engage internationally and across perspectives,” says CIFAR Director of Public Policy Brent Barron.
“If we look at what’s going to help the world respond in a way that best takes advantage of this technology, it’s research capacity, but also policy innovation and business leadership. We need to be able to bring all of those sectors together,” he says. Two lines of activities will serve this goal.
A call for workshops seeks proposals from around the world in three areas: policy implications; global dynamics; and creativity and discovery. These themes were selected to open the door to greater involvement from researchers in the social sciences and humanities.
“The vast majority of the people I’ve talked to are really interested in engaging across disciplines and across borders. We aren’t seeing many other mechanisms that are focused on supporting that kind of engagement.”
...we have great partners on the ground and a shared vision across the country. The Pan-Canadian AI Strategy is off to an amazing start
The workshops will build connections, and will also provide a pipeline for knowledge mobilization. Conclusions will be published as white papers, briefing documents and reports, for example, aimed at the government community, professional associations, NGOs and beyond.
The second part of the AI & Society program is a national series of policy foresight labs that will bring early career professionals in government, together with people from think tanks, advocacy groups and the broader public sector, to explore policy issues posed by AI. The aim is to spark discussions about the potential of technology, across policy domains.
As Barron sees it, Canada has all the pieces in place to contribute to the world’s moral compass on AI.
“If you look at everything that’s happening in Montreal, for example, on ethics and social responsibility, together with the Government of Canada’s investment in CIFAR on AI & Society, it seems as if Canada will be able to fulfil its traditional middle power influence in the world. As a global convenor and peacekeeper that exemplifies the values of diversity and inclusivity, this could be Canada’s time to shine.”
The scope of CIFAR’s AI efforts is necessarily ambitious. The overwhelming appetite for leadership and engagement on this issue, and the sheer number of players rushing into the field, makes the task of connecting everyone a challenge. Fortunately, CIFAR's model of convening diverse groups of actors to address broad questions of global importance has been refined and implemented successfully over 35 years.
“Because we’re very much in the early start-up phase of the strategy, outreach is really a critical component. One of the things that we are tracking very closely is all of the events and workshops and conferences and panels that we’re participating in, and our public visibility as the leaders of implementing the strategy,” says Strome.
So what is the vision for one year out from now?
“The AI Institutes will be buzzing with activity and undertaking leading-edge research to continue to advance AI technology,” says Strome. “We will have named a cohort of incredibly strong, diverse and internationally leading researchers who have come to Canada, or remained in Canada, and taken up roles as Canada CIFAR Chairs in AI. Our national program of activities will have reached a wider audience and brought experts from across disciplines and across the country together And, finally, our AI & Society program activities will have seeded new areas of research that help inform decision-making across a variety of sectors.
“It is ambitious,” acknowledges Strome, “but we have great partners on the ground and a shared vision across the country. The Pan-Canadian AI Strategy is off to an amazing start.”