A brain drain of artificial intelligence talent to the biggest tech companies threatens to set back academic research in the field, according to one of the three experts credited with breakthroughs a decade ago that lie behind today’s biggest advances in AI.

Yoshua Bengio, professor at the University of Montreal, was one of the pioneers of deep learning, a technique modelled on how the human brain works that has led to advances in language understanding and image recognition by computers.

“Industry has been recruiting a lot of talent — there’s a shortage in academia,” said Mr Bengio in a Financial Times interview. “It’s fine for those companies, but it’s not great for academia.”

Mr Bengio’s two counterparts have made the move into high-paying tech industry jobs, with Geoff Hinton from the University of Toronto joining Google and Yann LeCun, a student of Mr Hinton who moved to New York University, joining Facebook. Another early deep learning expert, Andrew Ng from Stanford University, is now chief scientist at Chinese internet search company Baidu.

Mr Bengio, whose brother, Samy, is a deep learning researcher who works at Google, said he had chosen to stay in academia to gain wider impact for his work. “I can contribute to all of humanity and not just the pockets of a single company,” he said.

Google announced on Monday it had made a C$4.5m donation to support AI research in Montreal, cementing the city’s standing as a centre in AI research. Thanks partly to Mr Bengio’s reputation, the University of Montreal and the city’s McGill University now claim 1,500 researchers working in the booming field, a concentration of talent that Mr Bengio said was greater than anywhere else in the world.

Other universities have lost some of their top AI talent as the biggest US tech companies race to build their own expertise. A year ago, Uber poached a team of researchers from Carnegie Mellon University in Pittsburgh, which had long been known as one of the leading centres for AI and robotics research. The ride-hailing company subsequently set up a research lab in Pittsburgh to work on developing driverless cars.

Google said it would set up a deep learning research centre of its own in Montreal to tap into work being done in the area. The search company has already assembled one of the biggest groups of AI researchers, much of it from academia, and last week added the head of Stanford’s AI lab, Fei-Fei Li, to its roster of experts.

Deep learning uses neural networks that emulate the layers of neurons in the brain. Despite attracting much of the excitement in recent AI research, Mr Bengio said the field was still at a rudimentary stage.

“People don’t realise the way deep learning is working right now is capturing very superficial aspects of our world,” he said. “Even a two-year-old can understand things a computer struggles with.”