Stephen Piron is co-founder of Dessa, a Toronto-based AI company formerly known as DeepLearni.ng
If you want to get an idea of the future of artificial intelligence, consider the internet mania of the late 1990s.
I studied computer science at the University of Toronto in 1999 and had an intimate view of the heady days of the internet’s hype. The internet and how it would change the world was on everyone’s minds and professors were always quick to remind us teenage undergrads that we’d be the rock stars of our time.
Today’s euphoria about AI feels eerily similar. As a co-founder of a Toronto-based AI company, I get weekly pitches from investment bankers looking to take us − a young company slightly more than two years old − public. While flattering, this attention seems premature. I also imagine we’re not the only AI company investment banks are calling on. These days, the world is hot for AI given its potential to transform our lives.
When it comes to the web’s first wave of hype, I’m sure most people recall how quickly the dot-com bubble burst. What the internet could deliver in those days wasn’t what people had been promised − the reality couldn’t surpass the hype.
This is something people should be thinking about more − a lot more − when it comes to AI, especially here in Canada. There’s more interest in Canadian tech than ever before, largely because of the country’s role in encouraging AI’s recent surge to the top. In 2012, Geoffrey Hinton and his students Ilya Sutskever and Alex Krizhevsky ushered in the current state of AI with their contributions to deep learning, the technology behind nearly all of the field’s recent advancements.
These days, the world’s biggest companies have been flocking to Canada to set up AI shops to cash in on these developments − from Samsung, which just launched a new lab in Toronto, to Alphabet’s DeepMind’s decision last year to set up a satellite location in Edmonton. Meanwhile, startup venture-capital financing from U.S. investors into AI and machine-learning companies globally has totalled US$321-million over the past three years, according to recent Thomson Reuters data.
Needless to say, there’s been a lot of hype so far about AI’s potential for the near term. While a lot of it has been noise, deep learning has been incredible so far in terms of delivering some very real and tangible impact. Some of its applications are innately fascinating: self-driving cars, artificially generated images and more or less instant language translations, to name a few.
Some of them have been less glamorous but also more reliable. The technology’s capabilities deliver your weekly Spotify playlist and are also helping call-centre employees with a better idea of why you’re calling them for help with your cellphone, before you actually make that call.
But when it comes to the potential that today’s leaders, investors and customers expect from AI, deep learning isn’t enough. If we’re going to truly realize what’s possible, we’re going to have to think bigger. It’s easy to forget, but deep learning is only a small part of the equation we’ll need to get right when we’re building AI systems that can thrive in already complex real-world environments.
Take major businesses for example: These are ideal environments that AI is being built for after its work in Big Tech and academia. Building AI that delivers more than just hype for enterprises requires technologists to think about a lot more than just writing code. Stakeholders, from programmers to the C-level executives, need to consider the existing networks of people, processes and tools that are already there.
So what’s the best way to forge ahead? From my perspective, the answer lies in thinking big, but starting small. Renowned systems thinker John Gall once wrote, “A complex system that works is invariably found to have evolved from a simple system that worked.” Mr. Gall’s got a good point: Until we get the basics right, we can’t move onto the things that will either meet or surpass our expectations.
This ultimately requires both the startups and big businesses developing AI systems to take a long-term view, swapping out the promise of shiny things for an appreciation of realistic timelines. Through collaboration, technology and business experts can keep expectations high while still iterating step-by-step toward an organization’s long-term AI vision.
While the great expectations for the internet’s early days far surpassed what was possible at the time, the promises made in the late 1990s eventually came true. It may have taken 20 years, and Pets.com may have been a commercial failure, but today, my mother literally buys kitty litter online at the click of a button.
Unlike many companies that fell into oblivion when the dot-com bubble popped, the companies that have stuck around all took a long-term attitude. Those in the AI business need to remember this.