Worried about AI? What the latest Canadian numbers say, and what actually helps

If AI makes you uneasy, you are not behind and you are not alone. In a 2025 survey of 1,025 Canadians, 79% said they were concerned about negative outcomes from AI, and only 34% said they were willing to trust it (KPMG and University of Melbourne, 2025). That is not a fringe view. It is most of the country, and very likely most of your staff.

It is also only half the picture. The same survey found 70% of Canadians expect AI to deliver real benefits, and 60% have already seen some. People are worried and curious at the same time, which is a sensible way to feel about a tool that is useful and still new.

What people are actually worried about

The worry is not one thing. When Leger asked Canadians in August 2025, 83% said they had privacy concerns about AI, and 83% said they fear society will become too dependent on it (Leger, 2025). In May 2026, 45% told Angus Reid they expect AI to significantly reduce the number of available jobs over the next decade (Angus Reid Institute, 2026). In the United States, where Pew asked workers directly, 52% said they were worried about how AI will be used in their workplace, against 36% who felt hopeful (Pew Research Center, 2025).

Those are fair things to worry about. Privacy is a real design question. Dependence is exactly what aviation spent decades learning to manage, which we wrote about in AI needs a pilot. And jobs do change when the work inside them changes. None of it is a reason to pretend the worry isn't there, in a staff meeting or anywhere else.

Leger found something else worth holding onto: 57% of Canadians had used an AI tool, up from 47% five months earlier, and 75% of those users rated the experience good or excellent. Worry tends to be highest before people have tried the thing on their own work.

What is happening inside Canadian businesses

The headlines move faster than most organizations do. In the second quarter of 2026, 19.2% of Canadian businesses were using AI to produce goods or deliver services. That is triple the share of two years earlier, and it still means about four in five businesses are not (Statistics Canada, 2026). Among the ones that aren't, the most common reason was not fear or cost. It was that 40% said AI is simply not relevant to their business.

Workers are often ahead of their employer's official plans. By March 2026, 35.9% of Canadian workers had used generative AI at work in the past year. The range is wide: 75.1% of people in management roles, 14.7% in trades and transport (Statistics Canada, 2026). So in a lot of organizations the question is not whether AI arrives. Someone on the team is already using it, with or without a plan.

On jobs specifically, small business owners are calmer than the public. Nearly 70% of CFIB members said they expect no impact from AI on their employment over the next 12 months, while nearly 45% already use generative AI in some way (CFIB, 2026).

What turns worry into confidence

This is the useful part. The research on what makes people comfortable with AI at work keeps pointing at the same few things, and almost none of them are about which tool you pick.

Leaders who use it themselves

BCG surveyed employees in 2025 and found the share who feel positive about generative AI rises from 15% to 55% with strong leadership support (BCG, AI at Work, 2025). That was the biggest single swing in anything we read for this post. Staff take their cue from whether the people above them use the tool on real work and say plainly what it is for, and what it is not for.

Training before rollout, not after

Only one in three employees in the same BCG study said they had been properly trained. In Canada, 24% of people have had any AI training at all, and 38% feel they have the skills to use it appropriately (KPMG and University of Melbourne, 2025). Worry and a lack of training travel together. The fix is rarely a course. It is a couple of hours with someone who knows the tool, on your own documents and your own tasks, before anything goes live.

Treating it as a people project

BCG's 2026 study of more than 600 US public companies puts a number on it: roughly 10% of the effort in AI that creates value is technology, 20% is algorithms and data, and 70% is people, processes, and organizational change (BCG, 2026). Change management research arrives at the same place from the other side. Projects with excellent change management meet or exceed their objectives 88% of the time. Where it was poor, 13% (Prosci, 2023).

Five things a small organization can do this quarter

• Name the task before the tool. "Answer after-hours inquiries the same night" is a project. "We should do something with AI" is a worry with a budget.
• Ask your staff what worries them, and write the answers down. Most of it will be specific and fixable: who sees the data, what happens to their job, who is accountable when it gets something wrong.
• Train on real work. Before go-live, not after.
• Keep a person signing off on anything that touches money, safety or your reputation. Then say so out loud. It is the most reassuring sentence you can offer a nervous team, because it is true.
• Write down how to switch it off. If nobody on the team can explain that in a sentence, you are not ready to switch it on. Our readiness checklist covers the rest.

A team that asks hard questions about a tool is usually the team that ends up using it well. The worry is not the obstacle. Treated seriously, it is the plan.