Facebook Heads to Canada for the Next Big AI Breakthrough
The first genuinely impressive AI assistant may well have a Canadian accent.
Facebook announced today that it is tapping into Canada’s impressive supply of artificial-intelligence talent and expertise by creating a major AI research center in Montreal. Several big recent advances in AI can be traced back to Canadian research labs, and Facebook is hoping that the new lab may help it take advantage of whatever comes next.
The new center will focus, in particular, on an area of AI known as reinforcement learning (see “10 Breakthrough Technologies 2017: Reinforcement Learning”).
The center will seek to apply this and other novel approaches to language, with the aim of producing more coherent and useful virtual assistants, says Yann LeCun, director of AI research at Facebook.
“Human children are very quick at learning human dialogue and learning common sense about the world,” LeCun says. “We think there is something we haven’t uncovered yet—some learning paradigm that we haven’t figured out. I personally think being able to crack this nut is one of the main obstacles to making real progress in AI.”
The lab will be led by Joelle Pineau, an associate professor at McGill University who specializes in applying reinforcement learning to language and robotics.
Facebook has already invested heavily in building smarter dialogue systems. While language remains a profound challenge for artificial intelligence, progress would pave the way for all sorts of new products and services. Despite serious limitations, voice assistants like Alexa and Siri, and text-based chatbots, have taken off in recent years. Others working on dialogue systems, including Apple, are also testing reinforcement learning (see “Siri May Get Smarter by Learning from Its Mistakes”).
Several leading figures in AI, including LeCun, have studied or taught at Canadian universities. Most notably, a technique known as deep learning, which has taken the tech world by storm in recent years, was honed by researchers working at the University of Toronto as well as other labs in Canada.
In deep learning, vast quantities of example data are used to train a large simulated neural network to perform a tricky task like recognizing a particular person in a photo or deciphering the words in speech. Deep learning has driven huge progress in speech recognition, online advertising, robots, and automated driving over the past few years.
Reinforcement learning builds on deep learning to let machines learn through experimentation. The approach has enabled robots to teach themselves new tasks through repetition and simulation. And researchers at the Alphabet subsidiary DeepMind used the technique to develop AlphaGo, a machine that taught itself to play the ancient board game Go at an expert level. David Silver, a researcher at DeepMind who led the development of AlphaGo, studied at the University of Alberta.
One challenge with using reinforcement learning for language, Pineau says, is that the task is open-ended, meaning there is no single, clear objective.
But experts say reinforcement learning could have a big impact in that area in coming years. “Most of reinforcement learning has been focused on robotics, so not as many tricks and modifications have yet been explored for language,” says Richard Socher, an adjunct professor at Stanford and the chief scientist at Salesforce. “That’s what makes it an exciting research area to explore.”
The location of Facebook’s new lab might be something of a warning sign for the United States. The Canadian government is investing millions in an effort to hold onto AI talent and foster new AI companies. A couple of months ago, DeepMind founded a lab at the University of Alberta, another center of excellence in reinforcement learning. Another AI-focused center was established at the University of Toronto with funding from the Canadian and Ontario governments, as well as a group of companies including Google. In contrast, the U.S. government has slashed funding for scientific research, including AI and machine learning.
The race for AI talent is, of course, already international. Facebook has a lab in Paris in addition to those in New York and Menlo Park, California. Several large Chinese companies, including Baidu and Tencent, now have U.S. AI labs, located in Silicon Valley and Seattle, respectively. And Google recently said it plans to open an AI research lab in China.
Moves by the U.S. government to restrict immigration could cause talent to gravitate toward other countries. “Immigration in Canada is more well organized, I would say,” says LeCun. “If you want to attract talent from outside the U.S., it’s actually quite a bit simpler to have this in Canada.”
Michael Bowling, a U.S.-born computer scientist who leads a lab at the University of Alberta that has produced cutting-edge poker-playing machines, says the new Facebook lab simply shows that Canada already leads the rest of the world in AI. “Researchers trained in Canada have been instrumental in many of the recent breakthroughs in AI,” Bowling says. “Canada is now gaining recognition as being an AI powerhouse, both in terms of innovative ideas and in terms of training the next generation of AI talent.”
Canada’s minister of science, Kirsty Duncan, said in a statement that the Canadian government has made strategic investments in artificial-intelligence research “because we are confident of its potential to create jobs and opportunities for Canadian scientists, engineers, and entrepreneurs.”
Indeed, after seeing AI researchers snapped up by big U.S. companies in recent years, Canada may well hope that the environment fostered by new labs, including the one in Montreal, will eventually produce companies that rival the likes of Facebook.
Geoffrey Hinton tells us why he’s now scared of the tech he helped build
“I have suddenly switched my views on whether these things are going to be more intelligent than us.”
Deep learning pioneer Geoffrey Hinton has quit Google
Hinton will be speaking at EmTech Digital on Wednesday.
Video: Geoffrey Hinton talks about the “existential threat” of AI
Watch Hinton speak with Will Douglas Heaven, MIT Technology Review’s senior editor for AI, at EmTech Digital.
Doctors have performed brain surgery on a fetus in one of the first operations of its kind
A baby girl who developed a life-threatening brain condition was successfully treated before she was born—and is now a healthy seven-week-old.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.