Startup leaders share insights about real-world tech adoption as AI grows
Canada has been a world leader in the development of artificial intelligence, but there is growing concern that the country is falling behind when it comes to adoption.
Recent market research by IBM showed that while 37 per cent of Canadian enterprises were using AI last fall, the country still lags behind the global rate of 42 per cent.
For companies that have yet to embrace this technology, the prospect of adopting AI may be daunting. To help guide new users through the process, four experts in the field provide their perspectives on what to keep in mind when approaching these tools in a business context.
Focus on a key outcome and work backward from there
“It’s very difficult to hire a team of data scientists and tell them to go ahead and build AI,” says Chloe Smith, the co-founder and CEO of Mercator, an AI-powered platform for the construction industry.
“They’re going to say ‘What are we trying to solve for? What data do we have?’ We can make the conversation more productive by asking: ‘Do we actually understand the different facets of what AI can do for us?’ ”
Smith suggests that business leaders need to first determine what they want to achieve. “AI is not the solution for most problems,” she says.
“We need to be clear about the data we have at our disposal and the underlying business question we are answering,” she says. “And once we get that answer, what would we do with that answer?”
Be proactive in identifying potential risks — and figuring out how to manage them
When the technology is developing so quickly, it can be difficult to properly assess new tools.
Karthik Ramakrishnan, the cofounder and CEO of Armilla AI, which provides audits and risk analysis of AI systems, says demystifying prospective tools should be the first step in adoption — to know what it can and can’t do.
“As AI becomes more pervasive, one of the biggest challenges — particularly for large organizations in regulated industries — is how to understand the risks of these models,” he says.
It’s important to ensure that not only will the model do what you want it to do, but where the failure modes could be, and to protect yourself accordingly.
“From our houses and cars to our ships and technology, insurance has always been a means to protect ourselves from residual risks,” says Ramakrishnan. “There’s a gap when it comes to the downsides of AI.”
Choose the most direct and efficient path to market
“If this is something that affects your main business product or service, what barriers are you facing and how can you get to the goal as quickly as possible?” says Patricia Thaine, the CEO of Private AI, which provides software to safeguard personal data. “Getting to market as quickly as possible makes all the difference.”
As Thaine explains, building a bespoke proprietary tool isn’t always the best option, especially for companies looking to compete on an international playing field. Working with partners or vendors who’ve already figured it out can expedite the process.
“What we have is a large group of companies who have experience trying to build things themselves over the past few years,” says Thaine.
“They’re going to start buying with a lens that’s much more informed, so they’ll be able to ask more informed questions of their vendors.”
Make sure everyone is speaking the same language
Mercator CEO Chloe Smith suggests that business leaders need to first determine what they want to achieve with artificial intelligence
Developing AI literacy is essential, says Mara Cairo, the product owner of advanced technology at the Alberta Machine Intelligence Institute.
She stresses the importance of ensuring that partners have a clear understanding of AI and large language model technology — which may require decoding acronyms and buzzwords and spelling out definitions.
“Let’s get everyone on the same page, even if you’re non-technical — especially if you’re non-technical.” Her recommendation is to ask some key questions: How do you measure success? What does success actually look like? What’s the benchmark you’re measuring against? And, perhaps most importantly: How do you determine that what you’ve built with AI is better than what you were doing before?
“AI is a really great tool for some really hard problems,” Cairo adds, “but we don’t want to use AI just for the sake of saying we’re using it — we want to be able to identify those problems.”