By CADE METZ
OSHUA BENGIO, ONE of the leading figures behind the rise of deep learning, is launching a Silicon Valley-style startup incubator dedicated to this enormously influential form of artificial intelligence.
The incubator, Element AI, will help build companies from AI research that emerges from the University of Montreal, where Bengio is a professor, and nearby McGill University, and he says this is just part of his efforts to develop an “AI ecosystem” in Montreal. Bengio says the Canadian city offers “the biggest concentration in the world” of academic researchers exploring deep learning, the breed of AI that now plays such an important role inside the likes of Google, Facebook, and Microsoft. “Element AI will help entrepreneurs get started in that high-growth area, with a team of experts—and my help—to steer those companies in the right direction,” he says.
According to Bengio, about 100 researchers are exploring deep learning at the University of Montreal and about 50 others are doing similar work at McGill. Whether this amounts to an unparalleled concentration of researchers is debatable—Europe is also a hotbed of research—but the importance of deep learning is undeniable.
Last week, Microsoft researchers unveiled a computing system that recognizes conversational speech as accurately as humans do. When transcribing audio from old telephone conversations, the system makes fewer errors than professional transcribers. This research carries caveats, but it underscores how much deep learning has accelerated AI in the last five years.
Using deep neural networks—networks of hardware and software that learn by analyzing vast amounts of data—several tech giants have pushed speech recognition to human-like levels, including Google, IBM, and Baidu. Using similar technology, these and other companies have built systems that accurately recognize faces and objects in photos. And now, deep learning is rapidly reinventing everything from machine translation to the Google search engine.
The deep learning movement grew out of a relatively small community of academics, and so many of these researchers have now funneled into the tech sector. One of the movement’s founding fathers, Geoff Hinton, works for Google. Another pioneer, Yann LeCun, runs the AI lab at Facebook. And in recent years, the giants of Silicon Valley have acquired one deep learning startup after another. Most notably, Google now owns DeepMind, the London-based think tank that recently used deep learning and other technologies to crack the ancient game of Go.
Hinton, LeCun, and Bengio are the three people most responsible for nurturing deep learning through 1980s, ’90s, and early 2000s when few others saw its potential. But Bengio hasn’t followed the other two into the private sector. Microsoft AI researcher XD Huang calls him “a great educator,” and that’s typically how the deep learning community describes him. That said, Bengio is an advisor to IBM, and with Element.AI, he’s creating a pipeline through which AI research can flow into the commercial world.
This is significant because deep learning talent remains relatively rare. Silicon Valley’s biggest players are engaged in an arms race for such talent that extends well beyond the usual fight for engineers. The enormity of the arms race was highlighted last year when OpenAI, a startup funded by Elon Musk and others, lured several key AI researchers away from Google and Facebook.
It’s also important that Bengio is building an incubator—something along the lines of Y Combinator—because this can address the gap that often exists between expertise in AI and the know-how needed to build a company or product. “A lot of technologists, particularly in deep learning, are very skilled at constructing powerful algorithms but don’t always know how to use them to solve real world problems,” says Chris Nicholson, the founder of a deep learning startup called Skymind. “These very skilled technologists need a bit a guidance.”
For researchers given the right guidance, the market for their skills is enormous. Deep learning is now technology that every big company needs. And there are only so many researchers to go around.
Update: Though Yoshua Bengio originally referred to his incubator as Element.AI, the official name is now Element AI.