In March 2026, Anthropic published The Labor Market Impacts of AI — a research paper introducing a new way to measure how AI is actually affecting jobs. Rather than speculating about what AI could displace, the researchers built an "observed exposure" metric that combines theoretical LLM capabilities with real-world usage data from their own AI assistant, Claude.
The findings are nuanced and worth unpacking — especially for Canadians making decisions about education and career paths right now.
The gap between capability and reality
One of the paper's most striking findings is the enormous gap between what AI can theoretically do and what it is actually being used for in the workplace.
Take Computer and Mathematical occupations: 94% of tasks in this category are theoretically feasible for LLMs, but only 33% show significant real-world AI usage. For Office and Administrative roles, theoretical exposure is around 90%, yet actual adoption is dramatically lower.
This gap matters. It means that headlines about AI "replacing" entire job categories are premature — but it also means there is a large reservoir of potential disruption that hasn't materialized yet.
Source: Anthropic Economic Index, "The Labor Market Impacts of AI" (March 2026)
Which occupations are most exposed?
The research ranks occupations by their "observed coverage" — the share of tasks where AI is both theoretically capable and seeing real-world adoption:
- Computer Programmers: 75% coverage
- Customer Service Representatives: high coverage
- Data Entry Keyers: 67% coverage
- Financial Analysts: high coverage
At the other end of the spectrum, occupations with zero coverage include cooks, mechanics, lifeguards, bartenders, and dishwashers — roles that require physical presence and manual dexterity.
The pattern is clear: AI exposure is concentrated in white-collar, knowledge-work occupations. Workers in the most exposed roles tend to be older, more educated, higher-paid, and disproportionately female compared to those in zero-exposure roles.
The young worker signal
Perhaps the most concerning finding for Canadian students and recent graduates: the research found a 14% decline in job-finding rates for workers aged 22–25 in AI-exposed occupations since the release of ChatGPT in late 2022.
This doesn't mean young workers are being laid off en masse. Unemployment rates in exposed occupations haven't spiked. Instead, the effect appears in hiring patterns — fewer new positions being filled by entry-level workers. The researchers note this finding is "just barely statistically significant," but it aligns with a troubling possibility: companies may be using AI to handle tasks previously assigned to junior staff, reducing the pipeline of new hires without visibly displacing existing workers.
For students currently choosing programs and planning career paths, this is a signal worth paying attention to.
What this means for Canadian education
While Anthropic's research focuses on the U.S. labour market, the implications translate directly to Canada. Our occupational structure, classified through the National Occupational Classification (NOC) system, maps closely to the U.S. categories studied. The most exposed occupations — software developers, financial analysts, administrative roles — are just as prevalent in the Canadian workforce.
For students choosing programs:
- Fields with high AI exposure aren't necessarily bad choices — but graduates need to enter them with AI fluency, not just traditional skills. A computer science degree paired with deep understanding of AI-augmented workflows is very different from one that ignores the shift entirely.
- Purely technical programs may face pressure if the entry-level tasks that traditionally served as training grounds are increasingly automated. Programs that emphasize higher-order thinking, complex problem-solving, and human judgment alongside technical skills will produce more resilient graduates.
- Trades and hands-on professions (the zero-coverage occupations) remain largely unaffected by current AI capabilities. For students drawn to these fields, the employment outlook is reinforced rather than threatened by these trends.
For educational institutions:
- Curriculum adaptation is urgent. Institutions that integrate AI tools and AI-aware pedagogy across programs — not just in computer science — will better prepare graduates for the actual job market they'll enter.
- Co-op and work-integrated learning programs become even more valuable. If entry-level hiring is tightening in exposed fields, students who graduate with real work experience have a critical advantage.
- Career counselling services need to engage with this data. Generic advice about "growing fields" isn't sufficient when the growth dynamics within those fields are shifting.
What career planners should do now
-
Research specific occupations, not just fields. The Anthropic data shows exposure varies dramatically even within broad categories. Use tools like E&EO's career explorer to look at 3-year employment outlooks for specific NOC unit groups across Canadian regions.
-
Watch the hiring pipeline, not just unemployment. The absence of mass layoffs doesn't mean the market is stable. Pay attention to whether entry-level positions in your target field are growing, shrinking, or being restructured around AI.
-
Build AI fluency regardless of your field. The occupations with the highest coverage aren't disappearing — they're transforming. Workers who can effectively collaborate with AI tools will have a significant advantage over those who either ignore AI or are displaced by it.
-
Consider regional dynamics. AI adoption rates vary by region, industry concentration, and employer size. A career that's heavily AI-exposed in a major urban centre may look different in a smaller market. Explore regional outlooks to understand how your local labour market compares.
-
Don't panic — but don't be complacent. The research is clear that large-scale displacement hasn't happened yet. But the gap between theoretical capability and actual usage suggests we're still in the early innings. Planning for a career that remains relevant as that gap narrows is prudent, not alarmist.
Looking ahead
Anthropic's researchers explicitly designed their framework to detect disruption before it becomes widespread — a kind of early warning system for the labour market. They plan to revisit these analyses periodically as AI capabilities evolve.
For Canadians navigating education and career decisions, the takeaway isn't to avoid AI-exposed fields. It's to enter them with clear eyes: understanding which specific tasks are being transformed, how hiring patterns are shifting, and what skills will remain distinctly human.
We'll continue tracking these trends and their implications for Canadian employment outlooks. Use our career investigation tools to explore how specific occupations are rated across the country, and check back for future analysis as new data becomes available.
This article draws on findings from Anthropic's "The Labor Market Impacts of AI" (March 2026). All statistics and figures cited are from the original paper. Employment outlook ratings for Canadian occupations are sourced from ESDC and Job Bank Canada.