By Jonathan Woods and John Lorinc
Artificial intelligence (AI) makes human decision-making easier. AI-based applications can automate tasks, understand and mimic human speech, and process information faster than any person—all with the goal of better service. These Canadian AI startups are transforming business inside and out.
Enabling AI at the edge
Karim Ali’s timing couldn’t have been better. In 2012, the founder and CEO of Toronto-based Invision AI was a research fellow in the artificial intelligence lab at UC Berkeley. That same year, deep learning algorithms achieved breakthroughs in accuracy that kicked off AI’s shift toward real-world applications.
“It got to the point where [AI] was accurate enough to be of practical use,” Ali says. “One of the issues though—and there lay the opportunity—people were getting this accuracy by using a lot of computational power.”
That recognition planted the seed that, in 2017, became Invision AI. The amped-up computing power required for deep learning is typically delivered with expensive graphic processor units or cloud-based computing services. Ali’s goal was to lower that cost by enabling deep learning networks to run on Internet-of-Things (IoT) devices using only the computing power of a device’s off-the-shelf factory processor.
This proposition combines deep learning and edge computing, an equally hot topic in industrial technology. Deep learning runs on layers of artificial neural networks designed to independently learn through repeated recognition of patterns in data, promising faster and more accurate insights for business. Edge computing refers to the embedding of data-processing capabilities within industrial sensors and devices, enabling them to perform computations locally. Ali’s idea of achieving the performance of deep learning with the relatively inexpensive hardware inside connected devices is enticing.
Invision AI has started to validate some of its technology. One demonstration came in a proof-of-concept project commissioned by the Ontario Ministry of Transportation in early 2018. Invision AI built a camera system that used a deep learning algorithm to determine the number of people inside a vehicle in motion in real time. Ali says, “We were able to get high 90s [per cent] accuracy with that.”
Other projects include collision-avoidance systems and enhanced motion cameras with object detection and classification capabilities. But these use cases are only a teaser of the company’s ambitions to own the intersection of AI and IoT. Ultimately, Ali says, “We want to be the operating system for edge AI.”
Money talks—just ask a chat bot
They might not offer service with a smile, but with 24/7 availability and growing efficiency, AI-powered chat bots promise savings and productivity enhancements worth billions.
The way Nat Cartwright sees it, today’s bots only hint at what’s to come in online customer service. Cartwright, co-founder of Finn.ai, a company that specializes in AI-driven “conversational banking platforms,” points to the financial sector’s growing use of natural language processing systems for internal interactions. They take the friction out of communications between front-line staff and back-office personnel, she says, while helping tellers or other customer service personnel get the right information more efficiently. “You still need people, but there are efficiency gains as well,” says Cartwright.
Dozens of providers, from startups to tech giants, are rushing to develop smarter service applications in this burgeoning market. A recent PricewaterhouseCoopers survey found three in ten executives believe chatbots can take over a range of administrative or logistical tasks that get in the way of high-value activities like planning or strategy development. Soon enough, these AI-based tools will also start altering business models and disrupting client relationships.
MindBridge Analytics Inc. is helping move that process along. Its machine-learning algorithms verify transactions—a capability that will radically alter the corporate audit process by gradually eliminating the long-standing practice of sampling and reviewing financial statements of publicly traded corporations.
MindBridge’s co-founder Eli Fathi predicts that in five to seven years, 90 per cent of all audit functions will be carried out internally by AI-driven systems. As with chatbots, he says the true payoff will come when firms figure out how to apply the insights gleaned from the information flowing into these systems to their strategic planning. Such shifts will give clients of audit firms new insights, Fathi says, while helping the firms to increase “their capacity and value.”