The Canadian federal government recently announced three specialized institutes to strengthen its expertise in artificial intelligence, and build on its history of pioneering these new technologies.

It’s time for Canada to invest in developing artificial intelligence

Alan Bernstein, Pierre Boivin, and David McKay

The Globe and Mail, 26th March 2017

Call me skeptical, but I’ve been working in AI long enough to have seen this before. I’ve been through similar ventures as an AI researcher in the UK, opening and closing through a sequence of “AI winters”. Most notably, there was the UK’s Turing Institute (arguably the origin of AI was Alan Turing’s work at Manchester), which closed its doors in 1994.

As for me, I hid during the AI winter of the early 1990s. I worked for a tiny startup, Scientia Ltd., now the world’s leading timetabling company. We did AI, but we didn’t tell anyone because nobody would have bought our products. We could have gained greatly from government support, but the Turing Institute acted primarily as an academic research consultancy, competing with startups like mine, not supporting them.

Now don’t get me wrong, I like institutes. I’ve worked for them, and having groups to translate research findings into the economy is essential. Universities are not always great at that.

The problem with the Turing Institute was that it became a new research-first institution that competed with universities, rather than complementing them.

So, I’d urge the new Vector Institute to take a different model: don’t be a university. And above all, don’t become an additional funding source for university researchers. Become something new. Become a pipeline for enabling results to generate social and economic impact, not just another source of research outputs.