"Six in ten Canadian jobs are exposed to AI" makes a scary headline. It's also a fair summary of a real Statistics Canada study. The trouble is what people hear: six in ten jobs will disappear. That isn't what the research says, and the gap between the two matters a lot if you're planning a workforce or writing interview criteria.
This post walks through StatCan's estimates of jobs exposed to AI in Canada, explains the idea of complementarity, and shows how to turn "this role is changing" into better hiring decisions.
#What did StatCan's AI job-exposure study find?
In September 2024, Statistics Canada researchers Tahsin Mehdi and René Morissette published Experimental Estimates of Potential Artificial Intelligence Occupational Exposure in Canada. They used two measures:
- AI occupational exposure: how much the abilities an occupation relies on overlap with what AI applications can do.
- Complementarity: how likely AI is to complement human workers in that occupation, based on job features such as communication, responsibility, physical conditions, how critical decisions are and how much preparation the job requires.
Combining the two, they sorted Canadian employees (May 2021) into three groups:
| Group | Share of employees | Approx. workers | Example occupations |
|---|---|---|---|
| High exposure, low complementarity | 31% | 4.2 million | Office support, administrative roles in finance and insurance, business and finance professionals, computer and information systems professionals |
| High exposure, high complementarity | 29% | 3.9 million | Nursing, education, health professionals, engineers (e.g. family physicians, secondary school teachers, electrical engineers) |
| Low exposure | 40% | 5.4 million | Transport and heavy equipment operators, construction trades, labourers (e.g. carpenters, welders, plumbers, firefighters) |
The 2016 distribution looked very similar, so this isn't a snapshot of a strange year.
#Does exposure to AI mean job loss in Canada?
No, and the authors are explicit. Their estimates are about technological feasibility: could AI, in principle, do things this job involves? That's a very different question from whether an employer will actually cut the role. As they put it, "exposure to AI does not necessarily imply a risk of job loss."
Real-world data so far lines up with that caution. StatCan's 2025 business survey found that 89.4% of AI-using businesses reported no change in employment as a result of AI, and its 2026 analysis says the employment impact "remains unclear," even as adoption rose to 19.2% of businesses. (We dig into those adoption numbers in what AI adoption in Canadian businesses means for hiring.)
#What is complementarity, and why does it matter?
Complementarity is the difference between "AI can do parts of this job" and "AI can do this job." In high-complementarity occupations, the work involves things like high-stakes decisions, direct responsibility for people and a lot of face-to-face communication. AI may handle some tasks, but a person still needs to be accountable for the outcome.
Imagine a hospital network in Ottawa. Its nurses are highly exposed: AI can summarize charts, flag drug interactions and draft discharge notes. But nursing is also highly complementary, because the nurse is the one who notices that the patient looks worse than the chart says and acts on it. AI changes the job, while the need for the nurse stays.
Now imagine the same network's records office. The roles are also highly exposed, but with lower complementarity: much of the work is structured, routine and done at a desk. Here, AI may genuinely reshape how many hours of which tasks are needed. That still isn't automatic job loss, but it's where planning attention should go first.
#Who is most exposed to AI in Canada?
A few patterns from the study stand out:
- Education: Unlike earlier waves of automation, AI exposure rises with education. 37% of bachelor's degree holders were in high-exposure, low-complementarity jobs, versus 25% of those with high school or less. Degree holders were also more likely to be in highly complementary jobs (46% for bachelor's, 58% for graduate degrees). And 73% of apprenticeship or trades certificate holders were in low-exposure jobs.
- Sex: 38% of women were in high-exposure, low-complementarity jobs, compared with slightly over one-fifth of men. Women (33%) were also more likely than men (25%) to be in highly complementary roles, partly because men are more concentrated in the skilled trades.
- Income: Exposure rises with earnings. The top income decile had the highest share of highly complementary jobs (55%).
For workforce planners, the education finding is the surprise. The roles most worth reviewing aren't on the factory floor. They're in offices.
#How should changing roles change your interview criteria?
Here's the practical part. If a role is shifting, the competencies you hire for should shift too, and your interview should measure the new ones consistently. Below is a starting map by exposure group.
| Exposure group | What's changing | Interview criteria to add or reweight | Criteria to de-emphasize |
|---|---|---|---|
| High exposure, low complementarity (e.g. office support, finance admin) | Routine drafting, data entry and lookups move to tools | Checking AI output for errors, handling exceptions, process improvement, clear escalation | Speed at routine tasks the tool now does |
| High exposure, high complementarity (e.g. nursing, teaching, engineering) | AI assists with analysis and documentation | Professional judgement, explaining decisions, communication under pressure, knowing when to override a tool | Memorization of information a tool can retrieve |
| Low exposure (e.g. trades, transport) | Little direct change for now | Safety, craft skill, reliability; light digital fluency for scheduling and reporting | Don't over-invest in AI questions |
#Turn each criterion into a structured question
Vague criteria produce vague interviews. "Good with AI" isn't measurable. These are:
- Verification: "Tell me about a time a report, tool or colleague gave you information that turned out to be wrong. How did you catch it, and what did you do?"
- Exception handling: "Describe a case that didn't fit the normal process. How did you decide what to do?"
- Judgement and override: "Tell me about a time you disagreed with a recommendation from a system or a senior colleague. What happened?"
- Learning agility: "What's a tool or process you had to learn quickly? How did you get up to speed?"
Then write a rubric for each one: what does a strong, acceptable and weak answer look like? Our guide to writing a structured interview scoring rubric covers that step by step.
#What does this look like in practice?
Imagine a Mississauga contact centre whose agents now get AI-suggested replies for routine customer questions. The role sits close to the high-exposure end. Rather than cutting agents, the centre decides its agents should spend more time on complex and emotional calls.
So it changes three things. It drops the typing-speed test. It adds two structured questions on de-escalation and on spotting when a suggested reply is wrong. And it reweights the rubric so judgement and empathy count for more than script adherence. The job title didn't change, but what "good" looks like did, and the interview now measures it.
That's the kind of change MSZ Hire is designed to support: structured voice interviews with live captions, where every candidate gets the same job-related questions, answers are scored against your updated rubric with quoted evidence, and a person makes every decision. There's no facial, voice-tone or emotion analysis, only what the candidate actually said.
#The bottom line on jobs exposed to AI in Canada
Six in ten Canadian workers are in jobs highly exposed to AI. About half of those are in roles where AI is more likely to help than to replace. And the evidence so far shows businesses adopting AI without large changes in headcount. "Exposed" is a prompt to rethink the work, not a forecast of layoffs.
For hiring teams, the response is concrete: identify which roles are changing, rewrite what "good" looks like, and interview for it consistently. Curious how a structured interview built around those criteria feels to a candidate? Try the candidate experience yourself. And when your roles evolve, MSZ Hire can help your interviews keep pace, with your rubric, your criteria and your people making the decisions.