Turning AI into a Competitive Advantage

The blistering pace of recent AI advancements makes it easy for leaders to feel they’re falling behind.

The Government of Canada’s recent AI Strategy dubbed AI as an increasingly ubiquitous part of daily life and pointed to a future increasingly dominated by the technology: “What we are seeing today is only the beginning.”

Business leaders of all stripes are prognosticating AI revolutions in everything from entrepreneurship to environmentalism to employment.

Companies large and small are doubling down on AI investment, shifting hiring forecasts and enlisting AI agents to autonomously accomplish tasks with a breadth and speed that no human could possibly replicate — much less a human who needs eight hours of sleep, enjoys unplugging on weekends and vacations, and catches a flu bug now and then. 

Amid all this noise, it can be difficult for leaders to get their bearings — let alone formulate strategies that prepare their organizations to thrive in this period of uniquely concentrated technological transformation. Smith School of Business professor Ceren Kolsarici likens it to feeling “like the ground is shifting under your feet.”

In a recent Smith Business Insight webinar, The AI Playbook: Lead, Compete or Be Left Behind, Kolsarici delivered a primer on how leaders can triage the ever-shifting opportunities and risks of AI, and plan to thrive in an increasingly autonomous future. 

Read on for three takeaways.

Takeaway #1: AI is not a magic wand 

The hype cycle can make it seem as though AI is both a panacea for all business challenges and a shortcut to maximize all commercial opportunities. The truth is a little less flashy, according to Kolsarici. “AI is, at its core, just a prediction machine at scale,” she said. “It takes the data that your business collects and learns patterns from it so it can predict what comes next. It’s not magic, and it’s really not mysterious.”

As algorithms and agents become more sophisticated, it’s very important for leaders to be cognizant of AI’s fundamental limitations, Kolsarici continued. “AI doesn’t create wisdom,” she said. “Even the smartest AI models now, those with reasoning logic, are still merely predicting.”

That’s why Kolsarici cautions against entirely handing over the reins to automated options. “In successful firms and successful brands, there is always a human initiating every action that AI completes and augmenting AI’s tasks to translate them into business decisions.”

Takeaway #2: Leaders should hyper-focus on value

The organizations that are harnessing AI most successfully today have one thing in common, Kolsarici said. It’s not the technical capabilities they use, nor the speed of deployment. Rather, it’s the ability of leaders to tune out distractions and zero in on what it all means. “They don’t start by asking what the technology can do,” she reasoned. “They start with the question: ‘Where can AI create value?’” 

In Kolsarici’s view, every AI initiative should link “very clearly and very deterministically” to a measurable business goal, and it’s the responsibility of leaders to both establish that focus and ensure everyone in the organization understands it. 

And that applies to more than just monetary value. “At every stage of design and deployment, leaders need to ask: ‘Who benefits from this system, and are those benefits distributed equitably?’” Kolsarici said. “Beneficence is the difference between building an AI that works and building an AI that works for people.”

Takeaway #3: Accountability is a differentiator

In 2022, an Air Canada chatbot gave customer Jake Moffatt incorrect information about the process to receive its bereavement rate for a round-trip flight to attend his grandmother’s funeral. When Moffatt followed the advice of the bot — to submit a claim for reimbursement within 90 days of the flight — the airline denied his request, citing its policy of granting bereavement fares at the time of booking, not retroactively. In the subsequent British Columbia Civil Resolution Tribunal case to resolve the matter, Air Canada argued that its chatbot was a separate legal entity responsible for its own decisions. The tribunal disagreed, siding in favour of the customer.

“Air Canada did not intentionally mislead a grieving customer,” Kolsarici explained. “But they deployed an AI tool without adequate governance, without testing edge cases, and without a clear accountability structure … and they ended up in court arguing their own software was not their responsibility.” This points to the growing disparity between deploying AI and governing it with accountability, Kolsarici continued — “a dangerous gap” in most organizations today. She pointed to research showing that while nearly all executives say AI ethics are important, most do not have formal AI ethics policies or gather metrics about risk, and most cannot explain how their AI makes decisions.

Organizations tend to focus too much on AI deployment, Kolsarici explained, and too little on what happens before (e.g. setting goals and checking for biases) and after (e.g. reviewing output and allowing humans to override it). Some tweaks to workflow architecture can ensure human safeguards are part of any AI use, she said, and create the kind of governance safeguards that deter situations like rogue chatbots. “We have to be able to govern outcomes and track impact.”