Back in the late 1980s, Geoffrey Hinton, who had already been teaching at Carnegie Mellon University in Pittsburgh for a few years, began to feel increasingly dissatisfied with the state of affairs in the U.S. He was particularly critical of Ronald Reagan’s foreign policy, especially the mining of Nicaraguan harbors, alongside concerns about the military funding behind the A.I. research he was conducting. When an opportunity arose to relocate to Canada, Hinton eagerly accepted.

“My wife and I were quite disillusioned with the U.S.,” Hinton shared with Observer, “and Canada seemed appealing.” Lured by a strong social framework and an attractive position as a fellow at the Canadian Institute for Advanced Research (CIFAR), Hinton made his way to Toronto in 1987. He has largely remained in Canada ever since, earning a Nobel Prize for his groundbreaking contributions to A.I. research.

Hinton’s move was not an isolated incident. Over the years, substantial funding for exploratory research has attracted numerous pioneering A.I. researchers to Canada, where significant breakthroughs laid the groundwork for the A.I. technologies that now dominate the tech landscape. Canada capitalized on this momentum in 2017 by implementing a national A.I. strategy, establishing three key A.I. hubs in Toronto, Montreal, and Edmonton.

Despite these contributions, many argue that Canada has not fully benefited from its own innovations. It’s not just ideas that have crossed the border to the U.S.; much of the country’s talent has followed suit. “Historically, Canada has faced the challenge of being a source of innovation without reaping the commercial benefits at home,” remarked Cam Linke, head of the Alberta Machine Intelligence Institute (Amii), in an interview with Observer.

While the establishment of competitive A.I. firms in Canada has often fallen short, a combination of increased government investment, strengthened research institutions, and shifting cultural perspectives is beginning to make a difference. For instance, the Toronto-based startup Cohere recently secured $500 million—an unprecedented figure for a Canadian generative A.I. startup—thanks to a mix of domestic and international investors. While acknowledging that Canada’s A.I. “brain drain” remains a challenge, Cohere co-founder Nick Frosst expressed optimism, stating, “I sense that the tide is changing.”

Drawing in Top Talent in A.I.

Long before the likes of OpenAI and Anthropic emerged, Canada had already established itself as a hub for ambitious A.I. research. Although it may not have matched the U.S. in terms of national funding, Canada became a haven for researchers interested in long-term experimental projects. Hinton recalls that three key figures—Rich Sutton, Yoshua Bengio, and himself—thrived in Canada due to its supportive social system and funding for foundational research.

Once Hinton settled at the University of Toronto in the late 1980s, Sutton, disenchanted with U.S. politics, moved to the University of Alberta, while deep learning innovator Bengio returned to Montreal to work at the University of Montreal. The presence of these influential researchers, coupled with Canada’s more favorable immigration policies, attracted numerous A.I. experts to the country, according to Amii’s Linke. “This dynamic created a cycle of talented individuals wanting to collaborate with these pioneers,” he noted.

Although geographically dispersed, Hinton, Sutton, and Bengio shared a deep passion for a specific area of A.I. research that had long been considered separate from traditional A.I. “Back then, there were two opposing camps: traditional A.I. focused on symbolic reasoning, while neural networks sought to emulate the human brain,” explained Hinton.

Despite being dismissed as “crazy theories” at the time, the neural network field received support from CIFAR. In 2004, CIFAR initiated the “Neural Computation and Adaptive Perception” program, which was directed by Hinton and included contributions from Bengio and LeCun. “It took time for neural networks to demonstrate practical applications, necessitating funding for researchers working on them without immediate, tangible results,” Hinton elaborated. “Securing funding in the U.S. was significantly more challenging.”

Researchers involved in this program convened annually to share their ideas, as noted by Ruslan Salakhutdinov, a professor at Carnegie Mellon University. In 2005, Salakhutdinov, then working in banking, unexpectedly encountered Hinton on the street. Hinton’s enthusiasm for deep learning models inspired Salakhutdinov to return to academia and pursue a Ph.D. under his mentorship.

Excitement in Canada’s neural net community surged in the early 2010s as advancements in neural networks, particularly in speech recognition, became evident. Hinton and his students, Krizhevsky and Sutskever, gained widespread attention in 2012 by winning an object recognition competition with their neural networks. This success led to the formation of DNNresearch, which Google later acquired for $44 million.

As neural networks gained traction, prominent researchers such as Hinton, Sutton, Bengio, and LeCun received enticing offers from tech giants like Google, DeepMind, and Meta. Many aspiring researchers from Canada opted to relocate to the U.S. in search of better opportunities fueled by lucrative salaries.

To counteract the brain drain in the A.I. sector, the Canadian government launched the Pan-Canadian A.I. Strategy in 2017, investing billions into A.I. research and establishing three A.I. hubs led by figures like Bengio, Sutton, and Hinton.

Despite the progress in A.I. research, the Canadian tech industry has been slow to adopt these innovations. Companies like BlackBerry and Element AI struggled to leverage the promise of neural networks due to conservative mindsets and financial hurdles. Additionally, the University of Toronto faced challenges in supporting entrepreneurial initiatives within its academic environment. Students leveraging university assets to transform research into startups often found themselves relinquishing larger stakes in their companies compared to their peers at prestigious American institutions like Stanford and Carnegie Mellon, as noted by Salakhutdinov. The University of Toronto indicated that equity agreements are typically negotiated individually, with the institution’s share varying from single digits to low double digits.

Moreover, Canada faces a significant challenge regarding computational resources compared to the U.S., which Hinton has termed the country’s “primary drawback” for emerging researchers. He recounted a situation where former student Jimmy Ba struggled to access the necessary graphics processing units (GPUs) for training large language models while at the Vector Institute. Ba eventually joined Elon Musk’s A.I. startup xAI. Hinton expressed concerns that Canada may face difficulties in achieving global leadership in A.I. due to resource limitations, despite its historical successes in foundational research.

While there has indeed been a talent exodus from Canada, some researchers have chosen to stay. Additionally, foreign companies have set up research facilities in Canada, creating opportunities for local graduates. Lacoste-Julien, who manages a Samsung lab at Mila, acknowledged the positive influence of these international offices in retaining talent after graduation. He pointed out that while the brain drain challenge is not entirely resolved, there have been notable strides forward.

The cultural values prevalent in Canada, which initially attracted luminaries like Hinton and Sutton, may also present challenges in competing with foreign entities in the A.I. sector. In provinces like Quebec, there exists a strong emphasis on quality of life and equality, often at odds with traditional success metrics tied to company size. Nonetheless, some emerging Canadian startups are challenging this narrative. For instance, Artificial Agency, co-founded by former Google DeepMind researchers, secured $16 million in funding this year for its innovative approach to enhancing gaming experiences through generative A.I.

The Canadian startup ecosystem is witnessing a resurgence, particularly in cities like Toronto. In 2022, the Canadian A.I. sector attracted $8.6 billion in venture capital, positioning the nation as a prime destination for A.I. investment. Companies such as Waabi, which focuses on autonomous vehicles, and Cohere, a rising player in the A.I. arena, have garnered substantial funding and attention. The growing support from firms like Radical Ventures, a leading Toronto-based venture capital firm, highlights the increasing potential of Canada’s A.I. ecosystem.

Collaborative efforts between businesses and research institutions in Canada, like Vector and Amii, have further propelled the growth of startups in the country. The retention of talent, especially in provinces like Ontario, reflects a shift in perspective among local businesses that now prioritize fostering growth within Canada. The experience of Artificial Agency underscores the evolving relationship between startups and academia, emphasizing the importance of attracting and retaining top graduates.

As Canada’s A.I. landscape continues to develop, a growing tradition of successful startups is paving the way for future generations of researchers and entrepreneurs. The focus on cultivating a robust local A.I. industry offers optimism for a sustainable and innovative future in Canada’s tech sector.