In Machines Like Me, Ian McEwan imagines a world in which humanoids powered by artificial intelligence are on the verge of becoming mass consumer products, kind of like Tesla Inc.’s electric cars circa 2013.
The potential is exciting. Charlie, the protagonist, is a mediocre day trader who becomes rich after he lets the robotic flatmate he purchased take control of his desktop. The robots turn out to be buggy and things go awry, but maybe not so badly awry that the defects couldn’t be fixed in an updated version.
McEwan set his story in the 1980s, a way of saying the future was yesterday. Anyone skimming the Bloomberg News website on Feb. 11 would have seen stories about machines making more money than human stock-pickers in a study by Indiana University and Boston Dynamics Inc.s robot dog going on patrol for Norwegian oil producer Aker BP ASA. We’re on the verge of something big.
For now, Canada is in the vanguard of nations that will begin to shape this new world. We won’t set the tone. The United States and China are so far ahead in AI and associated technologies that they won’t be caught. But we’re safely in the second tier, along with France, Israel, the United Kingdom and some others. According to Tomi Poutanen, the AI pioneer who sold the startup he co-founded to Toronto-Dominion Bank in 2018, the objective of local academics, business leaders and governments should be establishing Canada at the head of the pack chasing the bronze medal.
“We need to recognize that there is something happening here,” Poutanen, who is now TD’s chief AI officer, said in an interview in Toronto last month.
I suppose TD would say it was doing its part by purchasing Poutanen’s company, Layer 6, and underwriting the work of a few dozen leading AI researchers. Others might express disappointment that a promising upstart was absorbed by the banking oligopoly, although Poutanen doesn’t seem to think that his new masters are hindering his work. Still, it seems unlikely that Canada will develop a dynamic industry if the banks and international behemoths such as Alphabet Inc. buy up all the talent.
“Innovation comes from research, but it’s not enough,” said Julien Billot, chief executive of Montreal-based Scale AI, one of five outfits selected to participate in the Justin Trudeau government’s $950-million “supercluster” program. “You have to have research, but then you have to have an industrial sector and a startup sector where all this innovation can come together, because if not, we are funding professors and researchers and their competencies are bought by Google or Facebook or the American or Chinese players, which we don’t want.”
The supercluster initiative probably deserves more attention than it has received since February 2018, when Navdeep Bains, the industry minister, announced the winners of the competition that attracted dozens of entries.
Ever since Silicon Valley became shorthand for world-conquering innovation, governments everywhere have been trying to reverse engineer technology hubs of their own. Everyone knows what’s needed: universities, entrepreneurs and money. But getting the right mix of each is hard; so hard that some say it’s foolish to even try. After all, no one set out to create Silicon Valley, it just happened.
The economic Darwinians who argue that all industrial policy is folly tend to overlook the role that lucrative defence contracts played — and continue to play — in the development of the U.S. technology industry. Few countries have a military budget large enough to create new industries, but the U.S. example nonetheless shows that public funding can be a catalyst for investment and economic growth. The trick is designing a program that offers acceptable odds that taxpayers will get a positive return on their investment.
For awhile, the gossip was that the superclusters were in disarray. By way of example, it took Scale AI 10 months to reach an agreement with the federal government on how it would deploy its $230-million share of the funding. Its initial call for projects went out in May 2019, more than a year after Bains unveiled the winners. Scale AI announced its first investments at the end of June, and revealed a second batch of 10 projects in January. In all, Scale AI has co-invested about $30 million in projects worth a combined $80 million.
“The reality is they were right, up until May 2019,” Billot said in an interview last month when I asked him about the negativity surrounding the program. “They are wrong now.”
In hindsight, administrative problems were inevitable, as the federal bureaucracy’s culture of risk aversion was always going to clash with technologists’ desire for speed and flexibility. The result, though, is a thorough vetting process that should reduce the risk of boondoggles. Billot and his staff first select the applications they like. Next, they must win approval from a panel of outside experts. Scale AI’s board of directors has the final say.
Billot said the selection of more than a dozen projects in six months shows that Scale AI has found a groove. “It was hard to begin, but I think people should forget about this and just look at reality,” he said. “We have a great pipeline of projects.”
The pipeline suggests that the first AI-powered humanoid will not be Canadian.
Scale AI’s winning pitch was that Canada could develop an edge in logistics and supply-chain management. Its biggest commitment to date is $12 million for a group led by Air Canada that thinks it can use AI to “enhance passenger demand forecasting and cargo operations.”
Lousy prose, but there’s reason to follow this story: It’s a bet on the country’s future.
Financial Post
Email: | Twitter: carmichaelkevin