» Canada first to adopt strategy for artificial intelligence
22.11.2018 - Natural Sciences Sector

Canada first to adopt strategy for artificial intelligence

© VectorInstitute, From left to right: Yoshua Bengio, Geoff Hinton and Rich Sutton, Canada's pioneering AI researchers.

Artificial intelligence (AI) is emerging as a research priority for a growing number of countries. In the following blog, Drs Elissa Strome, CIFAR’s Executive Director of the Pan-Canadian Artificial Intelligence Strategy, Naser Faruqui, Director of Technology and Innovation at the International Development Research Centre (IDRC), and Remi Quirion, Chief Scientist for the Province of Quebec, explain why Canada was first off the mark to adopt its own national strategy.

In March 2017, Canada was the first country in the world to announce a national strategy for artificial intelligence (AI), with a CAN$125 million investment over the next five years by the federal government.

Many factors inspired the Canadian government to act. Canada had the talent advantage but we needed to act quickly to maintain that lead. International demand for talent, especially from the USA, was putting Canada’s prior investments in AI research and talent development at risk. There was concern in both government and in the private sector that this brain drain would compromise Canada’s capacity to become early adopters of this disruptive new technology.

The Pan-Canadian AI Strategy reflected the recommendations of CIFAR-led consultations within the Canadian AI research community. It is a significant, focused investment designed to advance research and innovation in AI, attract and retain some of the world’s leading AI researchers, develop a deep talent pool of highly-qualified personnel, and bring together thought leaders from around the world to examine the broad societal implications of AI. The first objective of the Strategy was to build on the critical mass and decades of research leadership that existed in Edmonton, Montreal and Toronto and establish three new centres of excellence in AI research and innovation, with Canada’s pioneering AI researchers in leadership roles: the Alberta Machine Intelligence Institute (Amii) at the University of Alberta in Edmonton, where Rich Sutton is Scientific Advisor; the Montreal Institute for Learning Algorithms (Mila), founded by Yoshua Bengio at the Université de Montréal, which focuses on reinforcement learning; and the Vector Institute for Artificial Intelligence in Toronto, where the University of Toronto’s Geoff Hinton is Chief Scientific Advisor.

Key to the ability to pursue basic research in deep learning and reinforcement learning at these three universities over the last 15 years has been the long-term, sustained support from national and provincial scientific agencies, including the Natural Sciences and Engineering Research Council of Canada, the Fonds de recherche du Québec and CIFAR.

CIFAR’s very first research programme in 1982 focused on Artificial Intelligence, Robotics and Society, so it was ahead of its time in its quest to understand machine intelligence and what it might mean for humanity. In 2004, CIFAR launched Canada’s Learning in Machines and Brains Program, previously called Neural Computation and Adaptive Perception. The programme originally fell under the leadership of Geoff Hinton and is now co-directed by Yoshua Bengio and Yann Lecun from New York University and Facebook. Over many years, their work led to the development of some of the AI-based technologies that we now rely on every day, including voice and image recognition and machine translation.

It is this long history of supporting groundbreaking AI research and building extremely productive global research networks that led the Canadian government to ask CIFAR to develop and implement the Pan-Canadian Artificial Intelligence Strategy.

A catalyst for private sector investment

The federal government’s investment in the Pan-Canadian Artificial Intelligence Strategy has catalysed significant investments from other levels of government and from the private sector. The provincial governments of Ontario, Quebec and Alberta have committed, or are expected to commit, additional funding of CAN$50-80 million each to their respective AI Institutes and the private sector has contributed a total amount of more than CAN$100 million to these same institutes.

Together, Amii, Mila, Vector, CIFAR and their partner universities have spent the past 18 months recruiting and funding researchers, while training the next generation of Canadian research leaders in AI.

Over the same period, Canada has seen an explosion in the number of AI-based start-ups, which have sprung up across the country in innovation hubs such as Montreal, Toronto, Waterloo, Edmonton and Vancouver. At last count, there were more than 650 AI-based startups in Canada, many of which are developing products and services that have the potential to make a social, environmental and/or economic impact.

Advancing the innovation agenda

So far, the Pan-Canadian Artificial Intelligence Strategy has focused on advancing academic research and innovation. At the same time, the Canadian government has developed a series of initiatives to engage the private sector and advance the innovation agenda. One example is the Innovation Superclusters Initiative, a CAN$950 million investment in regional industrial superclusters. A total of five superclusters were named in 2018. Only one of them may be explicitly focused on leveraging AI (scale.ai) but all five have plans to integrate AI into their strategy.

The federal government is also getting ready to announce a new AI Advisory Council, a multistakeholder group responsible for bringing the academic and private sectors together to offer advice and guidance on Canada’s approach to AI innovation and commercialization.

These initiatives are highly complementary with the Pan-Canadian Artificial Intelligence Strategy, as they favour strong collaboration between industry and academia on AI.

Canada assumes leadership role on the societal implications of AI

While AI offers tremendous opportunities to benefit society, there are also concerns with respect to employment, privacy, security, democracy and ethics.

Canada is taking a leadership role in the international conversation striving to understand the societal implications of AI. In March this year, the Government of Quebec proposed creating an Organisation mondiale de l'intelligence artificielle (Omia), an intergovernmental organization dedicated to fostering consensus among member states on the standards and practices that must govern the applications of AI.

Later the same month, the Fonds de recherche du Québec, in partnership with the Ministry of Economy, Science and Innovation, launched a call for proposals to all Québec universities for the creation of an international observatory on the societal impact of AI and digital technologies. The observatory will be mandated to support forward-looking research, knowledge mobilization and public engagement around the societal implications of AI. The creation of the observatory builds upon a grassroots movement that took shape over the course of last year, with the organization of the Forum on the Socially Responsible Development of AI on 3 November 2017, which adopted the Montréal Declaration for a Responsible Development of Artificial Intelligence.

In June 2018, in advance of the G7 meetings in Charlevoix, Québec, Prime Minister Justin Trudeau and President Emmanuel Macron issued a joint Canada-France statement on AI. They committed to establishing an international study group which would convene experts across countries and sectors and provide a forum for sharing analysis and best practices and providing foresight and coordination capabilities. A task force to determine the scope, governance and implementation plan for this international study group is due to issue a report by the end of this year.

There has also been an explosion in training opportunities for young people interested in applying AI to advance social innovation, with programmes such as Mila and McGill University’s joint AI for Social Good Summer Lab and the University of British Columbia’s Data Science for Social Good Fellowship Program.

Importantly, the Pan-Canadian Artificial Intelligence Strategy includes a dedicated research programme on the societal, ethical and economic implications of AI. CIFAR is working with researchers and partners in Canada, France (CNRS) and the UK (UKRI) to explore these issues and synthesize current thinking on the challenges and opportunities raised by this powerful new technology.

An emerging area of interest: AI for human development

Canadians have a desire to see AI applied for the benefit of all humanity. One emerging area of interest is AI for human development. The University of Ottawa’s Institute for Science, Society and Policy has been examining the societal impact of responsible innovation in AI.

IDRC has focused on advanced computing to solve development challenges. Building on research and experience that highlight how best to use the internet and computing to improve education, healthcare and governance, IDRC has developed a white paper on AI and development. This agenda for action underscores the opportunity to use AI technologies to combat entrenched problems such as misinformation and rumours, which can lead to conflict, violence and potentially even genocide. One project, Una Hakika, uses people and machine learning to verify rumours in troubled hotspots, such as the Tana Delta in Kenya and Rakhine State in Myanmar. Another project in Sri Lanka employs big data and algorithms to map and predict the spread of infectious diseases, in order to contain outbreaks better. Partnerships between Canadian researchers and those in developing countries have been a key ingredient in these innovative projects.

The positive potential of AI for developing countries is exciting but AI also poses significant social, political and economic risks. Machine learning approaches are only as good as the data they are trained on and inherent biases and inequities within a dataset can be amplified through machine learning.

Automation and job loss owing to AI are critical concerns for countries whose economies are driven by citizen-led micro-economic activities such as agriculture and crafts. These risks are not unique to developing nations – witness the impact of AI misinformation campaigns on recent US elections – but they may be exacerbated by weak governance and regulatory capacity. IDRC is embarking on a programme to help developing countries build innovative AI for good, identify not only the benefits but also the risks of AI for themselves and regulate and govern AI technologies.

Canada is one of a growing number of countries that have adopted national AI strategies, or are working towards this goal. These national strategies for AI are building on past efforts to digitize industry , as the 2015 edition of the UNESCO Science Report recalls. It is likely that AI will figure prominently among national research priorities in the next UNESCO Science Report, due for release in November 2020.

<- Back to: All news
Back to top