Demystifying the Value of AI
Insight Canada is home to the third-largest artificial intelligence (AI) talent pool, including Element AI Co-founder and deep learning pioneer, Yoshua Bengio.
In fact, Montreal has the largest concentration of deep learning researchers, worldwide. Bengio created Mila, the machine learning laboratory at the University of Montreal, and had more scientific citations added to his work per day than any other computer scientist in the world in 2018. He has become a renowned leader in Canada’s AI economy and actively contributes his expertise to various organizations. We spoke with him to unpack Canada’s unique opportunity to position itself as a global industry leader in AI technology.
Mediaplanet: What is your opinion of the way that the media portrays AI? Do you think they do a good job of raising awareness of the value of AI and machine learning (ML)?
Yoshua Bengio: There is no uniform description of AI in the media. Though you can find a lot of hype on AI technology, many readers forget that we're still very far away from human-level AI and super intelligent robots. Some companies also have an interest in exaggerating the intelligence of current systems, and the media often reports on information coming from biased parties who are not AI researchers. There is an important need for quality AI education in the media for computer science students, entrepreneurs, and decision makers across many sectors of the economy.
MP: What kind of benefits can a company executive expect from AI implementation?
YB: Specialized tasks which are currently performed by people who do not require a broad understanding of the technology can be automated if enough of the appropriate data can be collected. ML can also be used to train predictive systems which are trained based on future outcomes (e.g. to predict demand, traffic, etc). Automation can, of course, reduce costs but it can also delegate boring and dangerous tasks to machines. The technology can also be used to design completely new products and services and thus open new markets and create major disruptions and growth in various sectors.
MP: What do we need to do to maintain Canada’s position as top industry leaders in AI technology innovation?
YB: Our stature in this respect is growing fast, in great part due to start-ups. Canada is clearly a scientific leader in AI but much more effort is needed to translate that into innovation and to help start-ups grow and support companies in adopting AI. We need to rapidly expand the talent pool, both via accelerated training at many levels and an aggressive immigration strategy to attract researchers, engineers, and entrepreneurs. Finally, it is important to improve the VC funding market to avoid exits leading to the transfer of ownership and intellectual property outside Canada.
MP: What will happen if Canadians fall behind in their digital transformation to AI? How will that impact the Canadian business economy as a whole?
YB: We will miss out on a major growth opportunity and our companies will be in danger of being displaced by foreign companies. We need Canada to be not just a consumer of AI, but also a producer of AI. We need to develop both AI adoption in traditional companies as well as develop the AI-first industry. If we only consume foreign-produced AI, our governments will not have much access to the wealth produced by the technology, which they need to counter the potentially negative effects of automation on the job market.
MP: What is your best piece of advice for enterprise executives starting to evaluate their digital transformation to AI?
YB: Educate yourself and your team about AI, and seek help from Canadian partners who have AI expertise. Give members of your technical teams the time to train themselves. There are many ML training opportunities, for example at various Canadian AI institutes as well as in books, tutorials, videos, online scientific publications, free software packages, and more. Also, work on an AI strategy, which very much depends on having a coherent and long-term data strategy, based on projected uses of AI. Since it can take years to collect enough of the right data and to properly train your staff, it’s better to start this early.