Fewer artificial intelligence patents have been filed in Canada each year since 2016, despite over $350 million in announced government investment and the creation of 185 new AI companies since January 2016.
Over the past decade Canada had the fifth most patents globally that included the words “artificial intelligence,” based on an analysis by The Logic of data from the World Intellectual Property Organization (WIPO). That rank is at risk of falling significantly. In 2009, there were 183 AI patents filed in Canada; by 2017, that number dropped 61 per cent to just 72 patents. So far in 2018, 52 patent applications have been filed.
Canada is the only jurisdiction among the top 10 by AI patents filed to see a decrease in the number of patents applied for every year between 2016 and 2018. Sixth-place Australia, for example, had 311 patents in 2017—that’s more than Canada had in the past four years combined.
Canada’s decline comes despite the creation of hundreds of AI startups in recent years; there are currently around 650 such companies in seven of Canada’s largest cities alone.
Gartner Insights forecast that global business value derived from artificial intelligence will total US$1.2 trillion in 2018, which it says is an increase of 70 per cent from 2017. AI-derived business value is forecast to reach US$3.9 trillion in 2022, according to figures released last April by the analyst firm.
“AI promises to be the most disruptive class of technologies during the next 10 years due to advances in computational power, volume, velocity and variety of data, as well as advances in deep neural networks,” said John-David Lovelock, Gartner research vice-president.
While not the only indicator of commercial success, the dramatic drop in patents is surprising given Canada’s AI research heavyweights. Canada, which has the third most AI researchers in the world, is the only jurisdiction among the top 10 by AI patents filed to see a decrease in the number of patents applied for every year between 2016 and 2018.
Canada is home to some of the most prominent researchers in AI, including Geoffrey Hinton, internationally-respected “godfather” of deep learning, who works at Google and the Toronto-based Vector Institute. That research success, however, isn’t translating into Canadian patents.
“We’re leaders in research but we have very few Canadian tech companies that are making money from using AI,” said Jim Hinton, a Waterloo intellectual property lawyer (no relation to Geoffrey Hinton). “There’s a lot of research and early stage companies, but they’re not filing patents and they’re not bringing in real revenue yet.”
The problem, according to Hinton, is that much of the government money spent on AI goes to foreign-controlled companies who retain ownership of patents, even when they’re filed by researchers working in Canada. The resulting revenue generated by those patents then goes to the companies’ countries of origin.
According to an analysis by Hinton, 53 per cent of machine learning patents originally owned by Canadian researchers have been re-assigned to foreign corporations. The Logic’s analysis shows that the top 10 corporate owners of patents filed in Canada containing the words “machine learning” include Microsoft, Facebook, Google and Accenture. None of the top 10 are Canadian firms.
In the past two years, Facebook, Google, Microsoft, Uber, Thales, Samsung, LG and Fujitsu have opened or significantly expanded AI labs in Canada. The researchers working at those labs have been busy filing patents, all of which are owned by their foreign parent companies. Microsoft alone obtained 43 AI patents in Canada in the past decade, the second most of any company. The top nine companies holding AI patents in Canada are all foreign-owned.
“We’re just sort of the factory or the branch plant that sees the first stage, and then the high value elements are taken elsewhere, and that is a problem,” said Myra Tawfik, a professor of intellectual property commercialization at the University of Windsor.
Patents filed in Canada allow the owners to protect others from copying their invention or using it without paying to do so in Canada. Patents filed in other jurisdictions protect the intellectual property there. However, many of Canada’s top AI researchers aren’t filing patents in the country. When they are filing patents, those aren’t owned by Canadian companies. Hinton has nine patents, all filed in the U.S. under Google’s name. Yoshua Bengio, co-founder of Montreal-based Element AI and scientific director at the Montréal Institute for Learning Algorithms (MILA), has four patents owned by AT&T.
Bengio said he would be careful not to conclude too much from the number of patents, since in the machine-learning field, “most of the new ideas are being developed in academia or even in major industrial labs without patents being produced. Instead protection arises from publications.”
“The large number of patents is mostly due to Big Tech companies like Microsoft, Google and Facebook whose head offices are in the US. Many smaller companies are using industrial secret[s] to protect their innovation,” Bengio wrote in an email to The Logic.
Geoffrey Hinton declined to comment through a Google spokesperson.
Patents including the words “artificial intelligence” published in any language in the World Intellectual Property Organization’s database as of Dec. 7, 2018, were included. Some AI patents may not yet have been translated into English, or may not include the word “artificial intelligence” which would result in the numbers included here being smaller. However, it isn’t just fewer “artificial intelligence” patents being filed in Canada. The country has similarly seen a decrease since 2016 for patents including the words “machine learning” and “neural networks.”
Prime Minister Justin Trudeau has spent much of the past three years touting Canada’s artificial intelligence bona fides. At a November 2017 conference with Eric Schmidt, the Alphabet chairman told Trudeau he is “enormously thankful to Canadians” for their advancements in AI technology.
“We now use it throughout our entire business and it’s a major driver of our corporate success,” said Schmidt.
Patents aren’t the only measure of success, and by many metrics, Canada has a very strong AI sector. The country has the third-most AI researchers in the world, many of whom founded companies recently—185 in the past three years, according to Crunchbase data. Element AI raised $137.5 million last year, which one of its funders, the National Bank of Canada, called the “largest Series A funding round for an artificial intelligence company in history.”
It’s not just Element AI raising large sums. In 2018, Toronto companies Integrate.ai and Rubikloud raised US$37 million and US$30 million, respectively. All told, 244 Canada-based AI companies have raised a combined US$210 million in the past three years, according to Crunchbase data analyzed by The Logic. However, few of those companies are filing patents in Canada. Rubikloud, for example, has filed two patents in Canada, one in the U.S., and five international ones through the Patent Cooperation Treaty.
Bengio and Hinton have spoken publicly about the importance of open access and collaboration in AI research. U.S.-owned companies Google, Amazon, Facebook and Microsoft have all published open-source machine-learning software, and companies industry-wide encourage their researchers to publish their breakthroughs in academic journals in an attempt to foster collaboration. Meanwhile, despite the pledge to open-access principles, those same companies are rushing to file AI patents. There were 145 U.S. patent filings that mentioned “machine learning” in 2010. That number jumped to 594 in 2016.
Google sister company Deepmind Technologies exemplifies this trend. In 2018, Richard Sutton, DeepMind’s lead researcher in Edmonton, published 10 papers, all of which are available for any member of the public to download. Meanwhile, the company has filed 90 patent applications, 60 in 2018 alone.
Canadian companies share the open-access commitment, but not the patent holdings. Bengio, for example, has published 85 open-access papers in 2018. Over the same period, Element AI, the company he co-founded, hasn’t filed a single patent, according to WIPO’s database.
That may change soon though. Christophe Coutelle, vice-president of marketing at Element AI, told The Logic: “We’re a two year old company. We are filing 45 patent applications at the moment. As you know this takes roughly 18 months.”
Charles Plant, senior fellow at the University of Toronto’s Impact Centre, said that the lack of patent holdings isn’t the largest issue for Canadian AI companies.
“We aren’t going to research our way to greatness; we’re going to become great through sales and marketing. Patents are a defensive strategy,” said Plant. “AI is a tool and the use of that tool does not necessitate the creation of intellectual property in all cases. Canadian companies could, in theory, have success using open source tech or licensing patents from other companies.”
The global AI race is well underway. The Chinese government has poured US$150 billion into its AI strategy, France US$1.7 billion and, the U.S. defence department reportedly spent approximately US$7.4 billion on AI and the fields that support it in 2017 alone. Last month, the German government announced about US$3.4 billion for AI research and development between now and 2025. Canada has invested just over $350 million in AI.
In spite of Canada’s funding deficit relative to other countries, there are some areas where Canada leads in the world. In March 2017, Canada became the first country to release a national AI strategy. It includes $125 million over five years to, among other things, increase the number of Canadian AI researchers and graduates and create three new research centres: the Alberta Machine Intelligence Institute in Edmonton, the Vector Institute in Toronto, and MILA in Montreal.
Share the full article!Send to a friend
Thanks for sharing!
You have shared 5 articles this month and reached the maximum amount of shares available.Close
Share the full article!
Share the full article with your friends. Recipients will be able to read the full text of the article after submitting their email address. They will not have access to other articles or subscriber benefits.
You have shared 0 article(s) this month and have 5 remaining.
That focus on research will do little to help Canadian companies get more patents, according to Karima Bawa, former chief legal officer at Research in Motion (now BlackBerry).
“Canada is great at funding world class research but rather poor at protecting the IP that comes from that research so it actually benefits Canadian companies,” said Bawa. “When you look at countries like China and the U.S., they have pretty proactive measures to try and support people that are inventors in filing patents,” she said. “We have almost nothing.”
Canada is also at the forefront when it comes to ethical AI. Last week, Trudeau announced an International Panel on Artificial Intelligence, which will act as a global body scrutinizing AI projects to ensure the technology is grounded in human rights. A few days earlier, the Université de Montréal released the Montreal Declaration of Responsible AI, based on consultations with 100 technologists and ethicists and designed to set out a moral framework for using technology, including in applications like social credit scores and facial recognition. Declaration signatories anticipate it will help shape provincial, federal, and international law.
Governments across Canada are hoping that much of that AI development, ethical or otherwise, will take place here. Last week, Trudeau announced $230 million in funding for SCALE.AI, an artificial intelligence supercluster based in Quebec that the government claims will create over 16,000 jobs and over $16.5 billion in GDP impact in the next 10 years.