🔬 Research summary by Dr. Cristian Gherhes (@cristiangherhes), Research Fellow at Oxford Brookes Business School.
[Original paper by Cristian Gherhes, Tim Vorley, Paul Vallance, Chay Brooks]
Overview: This paper explores the emergence of AI as a new industry in Montreal, Canada. It highlights the key roles that different actors (i.e., individuals/organisations/institutions) played individually and collectively over three decades in creating the thriving AI ecosystem that put Montreal on the world AI map.
Introduction
How do new industries come to life? And why do they grow in some places and not others? One view is that this has a lot to do with serendipity and the type of industrial activity already present in a region. But what about new, radical technologies like AI? How do such technologies become established and develop into new industries? Focusing on Montreal, this paper shows that the emergence of AI in the region cannot be attributed to historical accident but is the result of sustained work by a number of key actors who contributed to building an AI technological innovation system over three decades.
The paper refers to these actors as “system builders” and they include trailblazers (i.e., pioneering actors with a fundamental role in the development of AI in Montreal, like AI scientists, venture capital firms (VCs) and AI start-ups), anchors (i.e., universities, public research labs, and multinational enterprises who contribute to knowledge creation and talent development), and the state (federal, provincial, and local government actors). Each of these actors performed different roles, contributing in specific ways to the development and growth of the AI ecosystem. The paper highlights their role across two phases of system building: a long phase of scientific discovery and exploration, followed by a more strategic phase of intense system building. We show that AI did not just happen to Montreal; it took the efforts of multiple actors to build a thriving AI ecosystem.
Before the hype: the scientific exploration phase (early 1990s–2016)
It took more than two decades for the regional AI ecosystem to develop and put Montreal on the world AI map. This is because AI has not always been a hot topic, and this was especially the case in the 1980s and 1990s, a period also known as the second “AI winter”, when public interest in AI waned and funding dried up.
But all of this did not matter for a small group of researchers led by Prof Yoshua Bengio who carried on with their research on artificial neural networks when the world was ready to move on. Here is the catch though: they would not have been able to do so without the critical government funding that kept flowing to support their work. Key actors here are the state and the Canadian Institute for Advanced Research (CIFAR) in particular—a publicly funded organisation whose ethos of scientific curiosity and exploration promotes long-term fundamental research. Very importantly, the research grants it awards are not tied to commercial objectives, which enables researchers to tackle big scientific questions by working together over long periods of time. Luckily for Montreal, and for Canada, AI was one of the programmes funded by CIFAR that kept research in neural networks alive during the AI winter.
While impossible to predict at that point that AI will grow into the industry that it is today, public funding was critical to its development. It supported Prof Bengio’s work at the Montreal Institute for Learning Algorithms (MILA), which he founded in 1993, creating a small network of AI researchers. MILA ultimately played a key role in advancing the field of AI. Trailblazing through this period of uncertainty, researchers at the institute made major scientific contributions, including breakthroughs such as deep learning, curriculum learning, and generative adversarial networks (GANs) which underpin many of the AI innovations that we see, and use, today. They also contributed to building a strong academic pillar and to training the next generation of AI researchers—a critical resource for the ecosystem that will later fuel its growth.
However, this is not all. A hidden trailblazer—a VC firm—entered the nascent ecosystem in the late 2000s. On a mission to build an entrepreneurial ecosystem in Montreal, which back then had a weak entrepreneurial scene, the VC helped build the infrastructure that now supports start-up activity in AI and beyond. This involved launching accelerator programmes, building international links and bridges between industry, academia, and government, and starting much-needed seed funds. This early activity, while not directly linked to AI, promoted the development of an entrepreneurial culture and paved the way for growth.
Riding the AI hype: the strategic development phase (2016 onwards)
The scientific breakthroughs revived global interest in AI and propelled Montreal’s nascent ecosystem onto a more intensive system-building phase. Around 2016, AI activity in Montreal saw a boom, and it was an AI start-up that started it and put Montreal in the global spotlight. Founded in 2016, Element AI quickly became the fastest-growing AI start-up in the world, raising what was then considered record funding for an AI company. The fact that Prof Yoshua Bengio, who by this point became known as one of the godfathers of deep learning, was a co-founder, boosted the company’s, and the ecosystem’s, credibility. Its rapid success catalysed the growth of the AI ecosystem which became a magnet for international talent, VCs, and multinationals—all attracted by the concentration of AI expertise.
What followed was a wave of start-up activity and new actors entering the ecosystem. Very prominent among these are foreign tech giants like Microsoft, Google, and Facebook, who were among the first to open research labs in Montreal and to anchor themselves in the ecosystem by working with star academics, building links with universities and start-ups, and providing research funding—others soon followed suit. This gave credibility to the ecosystem and signalled its potential to investors, talent, and entrepreneurs both within Canada and internationally.
The AI hype that followed helped attract critical resources, particularly money and talent, and the actors that paved the way for growth in the early AI days became part and parcel of the growing AI ecosystem. The renowned AI researchers attracted more AI researchers and funding, which attracted more entrepreneurs and companies into AI, which attracted more VCs, and the impact compounded, turning Montreal into an AI hotspot. The Canadian Government also stepped up its role as an enabler through key strategic AI investments alongside the Government of Québec. These prominently include the CIFAR-led Pan-Canadian Artificial Intelligence Strategy and the Innovation Superclusters Initiative which saw CAD$230m invested in the Montreal-based SCALE.AI supercluster. Besides these, a range of tax incentives at both federal and provincial levels, investments in AI accelerators, and a friendly immigration policy have made it attractive for AI start-ups, multinationals, and talent to establish in the region.
The ambition is to make Montreal the Silicon Valley of AI. And it looks like Montreal has thought of everything. Just when concerns relating to the development and use of AI in the economy and society started to make the headlines, Montreal was already advocating for ethical and socially responsible AI through the Montreal Declaration for a Responsible Development of Artificial Intelligence. Other initiatives quickly followed, including the Montreal AI Ethics Institute and the publicly funded International Observatory on the Societal Impacts of Artificial Intelligence and Digital Technologies (OIISIAN). But the journey does not stop here. The key question is: is there more to this hype?
The evolution of the AI ecosystem over the three decades is summarised in the figure below.
Where next for Montreal?
The AI hype got Montreal dreaming of AI unicorns. It is the collective vision of those involved in AI in Montreal that the city becomes the home of the next tech giants. However, doing AI research is one thing, commercialising it is another. There are concerns that the scientific breakthroughs have created inflated expectations and that benefitting economically from AI is easier said than done. While no doubt Montreal has earned its status as a centre of excellence in AI research, questions remain over its ability to mature and generate the next generation of AI companies.
Just last year, something that many feared, happened. In November 2020, Element AI, Montreal’s AI poster child, announced its acquisition by American software company ServiceNow. This is far from the successful exit that everyone envisaged at the start, given that the company was sold for less than the total amount of capital raised—a significant loss for the investors, including the taxpayers. Beyond that, the exit raises important questions for the future of the AI ecosystem, which lost a key anchor firm. Will Montreal lose credibility? Will investors be more cautious? Will talent stick around? Will another AI winter come?
Some factors are beyond anyone’s control and only time will tell, but Montreal has built an incredible technological innovation system around AI, and those who helped build it plan to stay and to continue to build. With the highest concentration of academic researchers in deep learning in the world, a growing fundraising scene, more than $2bn in foreign direct investment in 2019, and new AI companies starting up or joining the ecosystem every year, there are reasons to be optimistic.