Blog
Your team already uses AI. Do they have a safe way to do it?
A new survey of Canadian workers puts numbers on something many business owners suspect. BetaKit reports on research from Edmonton's Punchcard Systems, done with Angus Reid, which surveyed 1,100 working Canadians in September. Nearly half of heavy-industry workers (49%) say they use AI at work. Of those using AI, more than two in three say they've had no formal AI training from their employer. In construction, that rises to 87%.
"In heavy industry, adoption is being driven by workers using a chatbot or general-purpose AI tools on their own, well before most companies have implemented formal AI tools and guidelines on how to use them throughout the business," Punchcard co-founder Sam Jenkins told BetaKit.
The survey is about construction, manufacturing, energy and similar fields. It's still worth asking the same question about your own team, whatever you do: are people already using AI at work, and has anyone told them how?
What workers trust AI with
The survey found workers mostly trust AI with paperwork, reading technical documents and finding how-to information. They stop short of trusting it with decisions that affect worker safety.
That's a sensible line, and it's a useful starting map. The jobs people already hand to AI on their own are the jobs where they feel the most friction. Those are often the best candidates for a proper tool.
The quiet risk: helpful people, public tools
When staff use AI on their own, they're usually trying to do their job better. The intent isn't the problem. The problem is where the information goes.
Picture a site supervisor pasting a client's drawings, a contract clause or an incident report into a free chatbot to get a quick summary. That data is now sitting with a service nobody at your company chose, under terms nobody at your company has read.
It isn't only a privacy question, either. If the answer that comes back is wrong, nobody has checked it, and it can end up in a report or a quote to a customer.
Banning AI outright tends to push that use out of sight rather than stop it. A clear, simple path usually works better.
Five steps to turn quiet AI use into a plan
1. Ask, without blame. Run a short, anonymous check-in: which AI tools are people using, and for what? You can't plan around use you can't see.
2. Write a one-page rule of thumb. Plain language. What's fine to paste into an AI tool, and what never is: client details, personal information, passwords, anything under a confidentiality agreement.
3. Start with the tasks people already trust AI with. Paperwork, looking things up in manuals, how-to questions. Those are low-risk, high-friction jobs and a good first project. Pick one, set a clear goal, such as cutting the time it takes to fill in a daily report, and learn from it before moving to the next.
4. Let the people who do the work shape the tools. Punchcard makes the same point: workers should help decide what AI at work looks like. They know where the time goes and where the mistakes happen.
5. Keep safety and big decisions with people. Your workers have already drawn that line themselves. Write it into your guidelines so it stays drawn as the tools get better.
None of this needs a big budget or a long project. It needs someone to own it and a few hours of honest conversation with your team.
Sources
- "Half of heavy industrial workers use AI, but workplace training is lacking", BetaKit, 2026-10-06. https://betakit.com/half-of-heavy-industrial-workers-use-ai-but-workplace-training-is-lacking/
- "Study: Over 2 in 3 Canadian Heavy Industry Workers Received No Formal AI Support From Their Employers", Punchcard Systems via Business Wire, 2026-10-06. morningstar.com (Business Wire)
- Punchcard Systems, "The State of AI at Work in Canada", with Angus Reid Group, fieldwork 2026-09-02 to 2026-09-11, n=1,100.
