From Digital Divide to AI Divide: The Next Equity Challenge
Canada spent two decades closing its broadband gap. A new gap is opening on top of it — measured not in megabits, but in who gets to use artificial intelligence at all, and who is left interpreting it from the outside.
August 2026
The gap in generative-AI use between the OECD's heaviest and lightest user groups by age — the widest divide the organisation has ever measured for any digital technology, wider than the gaps by income or education (each ~21 points) and far wider than the 4.2-point gap by gender. [7]
Executive summary
- Canada's older digital divide — access, speed, cost — is roughly 90% closed in cities but still runs deep in rural, remote and northern communities. [10]
- A second, faster-moving divide is opening on top of it: who actively uses AI. It tracks income and education almost exactly, and by age it is the widest gap the OECD has ever recorded for a digital technology. [7]
- Canadian business AI adoption tripled in two years (6.1% to 19.2%) — but the tripling happened on top of an urban-rural gap that also roughly tripled in absolute terms, and a 37.8-point spread between the most- and least-digitized sectors. [2]
- Unlike the broadband divide, the AI divide cannot be fixed with wires alone — it is a literacy, confidence and trust gap layered over an access gap, and 40% of Canadian businesses still say AI is simply "not relevant" to them. [2]
- Ottawa's 2026 "AI for All" strategy is the first federal attempt to treat AI equity as its own policy problem, distinct from broadband — with literacy programs explicitly routed through libraries and community organizations in rural, remote and northern regions. [9]
Data dashboard
of Canadian businesses used AI to produce goods or deliver services, Q2 2026 — up from 6.1% in Q2 2024
Statistics Canada
urban vs. rural business AI adoption in Canada, Q2 2026 — urban firms adopt at more than double the rural rate
Statistics Canada
of Canadians "very prepared" with workplace digital skills today — projected to fall within five years as technology outpaces training
ABC Life Literacy Canada, 2025
of Canadians had used a generative AI tool in the past year as of mid-2024, mostly to experiment
CIRA Internet Factbook 2024
of Canadian businesses say AI is "not relevant" to their operations at all — nearly as many as those using it
Statistics Canada, Q2 2026
Source: Statistics Canada, Canadian Survey on Business Conditions [2]
Source: Angus Reid Institute, May 2026, n=1,842 Canadian adults [8]
01 — The old divide isn't closed, it's a foundation
Canada never finished bridging the first digital divide
Before there was an AI divide, there was a broadband divide, and Canada is still living inside it. The country has "basically met" its national target of 50 Mbps download / 10 Mbps upload speeds in urban areas, but the 2024 Communications Monitoring Report puts rural households at roughly half the coverage rate of urban ones — a gap that the CRTC's own data shows has not moved since 2020. [10] The divide is sharpest in the North: sparse populations, extreme distances between communities and the physical cost of running fibre or satellite service through Arctic terrain have kept Northern Canada as the country's most persistently under-connected region for two decades. [11]
Layered onto geography is a skills problem that connectivity alone won't fix. Only 30% of Canadians currently rate themselves as "very prepared" with workplace digital skills — and literacy organizations project that number will fall to 23% within five years, not because skills are eroding but because the bar keeps rising faster than training keeps up. [1] That combination — uneven wires plus uneven confidence — is the foundation the AI divide is now being built on top of, and it means AI equity work cannot start from zero: it inherits every crack in the digital-inclusion effort that came before it.
02 — A faster, sharper divide
The AI gap is opening along the same lines — only wider
If the broadband divide took twenty years to narrow, the AI divide is forming in two. Statistics Canada and the Angus Reid Institute both find that AI use is not distributed evenly across the population that already has internet access — it tracks income and education almost exactly. Active use of AI tools (prompting them directly for information, not just encountering AI passively) rises from 21% among Canadians earning under $50,000 to 48% among those earning $200,000 or more, and from 23% among those with a high-school education or less to 41% among university graduates. [8] Those are the same Canadians who were already best positioned in the labour market before AI arrived — meaning the technology is, for now, amplifying an existing advantage rather than levelling one.
The starkest split, though, is by age. Across the OECD, the gap in generative-AI use between the most- and least-active age groups is 53.6 percentage points — nearly two and a half times the gap by income or education, and more than twelve times the gap by gender, which sits at just 4.2 points. [7] Three-quarters of students now use generative AI tools, and 41.1% of employed adults do, compared with only 12.5% of retired or otherwise economically inactive people. [7] That is not simply a preference gap — it is a structural one, because the people least exposed to AI today are disproportionately the people who already face the steepest re-entry costs if the labour market changes around them.
03 — The business side tells the same story
Adoption by Canadian firms is following the map, not flattening it
Canadian business AI adoption has genuinely accelerated — from 6.1% of firms in the second quarter of 2024 to 19.2% by the second quarter of 2026, a tripling in two years. [2] But the growth curve is not lifting all boats equally. Urban businesses used AI at 21.0% in Q2 2026 versus 9.9% for rural businesses — more than double, and importantly, the absolute gap has also widened even as both rates have tripled off their 2024 base. [2] Sector matters even more than geography: information and cultural industries sit at 42.3% adoption, while agriculture, forestry and fishing sit at just 4.5% — a 37.8-point spread that maps almost exactly onto which sectors were already the most digitized before generative AI existed. [2]
Perhaps the most telling figure is the one that looks like non-adoption but may be something else: 40% of Canadian businesses say AI is simply "not relevant" to them. [2] For some, that is likely true today. For others — a rural retailer, a small agricultural operation, a community-based nonprofit — "not relevant" may really mean "not accessible," a polite way of describing a gap in awareness, capital or connectivity rather than a genuine absence of use cases. Distinguishing those two groups is exactly the kind of diagnostic work Canada's AI equity strategy will need to do well.
At a glance
| Dividing line | Higher-use group | Lower-use group | Gap | Source |
|---|---|---|---|---|
| Age (individuals, OECD) | Students / working-age | Retired / inactive (12.5%) | 53.6 pp | OECD, 2026 |
| Sector (Canadian firms) | Information & culture (42.3%) | Agriculture / forestry / fishing (4.5%) | 37.8 pp | StatCan, Q2 2026 |
| Income (individuals, Canada) | $200k+ (48%) | <$50k (21%) | 27 pp | Angus Reid, 2026 |
| Education (individuals, Canada) | University (41%) | HS or less (23%) | 18 pp | Angus Reid, 2026 |
| Geography (Canadian firms) | Urban (21.0%) | Rural (9.9%) | 11.1 pp | StatCan, Q2 2026 |
| Gender (individuals, OECD) | Men | Women | 4.2 pp | OECD, 2026 |
04 — Why it compounds
This divide doesn't stay a technology story for long
Access and adoption gaps are early-warning signals, not endpoints. The same demographic groups showing the lowest AI engagement today — lower-income workers, workers without post-secondary credentials, older workers outside the labour force, and businesses in low-digitization sectors — are also disproportionately represented in the occupations that other research in this series finds most exposed to AI-driven disruption of entry-level hiring, and least likely to have access to employer-funded reskilling. In other words, the AI divide is not evenly distributed risk: the groups with the least practice using AI today are often the same groups with the most to lose if they are not fluent in it by the time it reshapes their sector.
This is also a competitiveness story, not only a fairness one. Canada already lags peer economies on AI diffusion — trailing markets like Ireland, New Zealand and Belgium despite strong research capacity — and an uneven domestic adoption base makes that harder to close, because national productivity gains from AI depend on adoption reaching well beyond the businesses and workers already primed to use it. [3]
05 — The federal response
"AI for All": Canada's first attempt to treat this as an equity problem
In 2026 the federal government launched AI for All, its national AI strategy, with an explicit goal of raising business AI adoption from roughly 12% to more than 50% by 2030 and generating up to 250,000 new jobs through AI adoption. [9] Unlike earlier digital-inclusion efforts, which centred almost entirely on physical connectivity, this strategy names AI literacy as its own deliverable: a National AI Literacy initiative aims to reach one million entry-level post-secondary students and train more than 3,000 educators with AI learning kits, while a modernized Job Bank (backed by $50 million over five years) will use AI-powered matching to connect people to opportunities that fit their existing skills profile. [9]
Notably, the strategy explicitly routes literacy delivery through public libraries and community organizations "especially those in rural, remote, and northern regions" — a direct acknowledgment that AI literacy cannot simply piggyback on institutions (colleges, large employers, urban school boards) that already reach the digitally-advantaged first. [9] A further $30 million through CanCode is earmarked for free K-12 digital and AI skills training aimed at underrepresented groups. Whether this reaches the households at 58% broadband access instead of 97% will be the real test of the strategy, not the headline adoption target.
Recommendations
- Track AI-adoption equity (by income, age, geography, sector) with the same rigour as broadband coverage — publish it alongside connectivity data, not separately.
- Fund literacy delivery through trusted local intermediaries (libraries, community organizations, Friendship Centres) rather than only through digitally-advantaged institutions.
- Treat the 40% of "AI not relevant" businesses as a segmentation problem — separate genuine non-use cases from access and awareness gaps before designing support.
- Build AI literacy into adult and continuing-education pathways, not only post-secondary — the steepest age gap sits outside the school system entirely.
- Pair digital-skills training with confidence-building, not just tool tutorials — the "very prepared" figure is falling despite rising tool availability.
- Prioritize outreach to lower-income and rural learners, who show the fastest willingness to adopt once given structured, low-cost entry points.
- Audit who inside the organization already uses AI daily versus never — internal adoption gaps usually mirror the external ones by role, tenure and location.
- Don't wait for sector peers: low-digitization sectors have the most to gain precisely because so little of the addressable productivity gain has been captured yet.
- Extend AI training to remote and rural staff on the same timeline as head-office staff, not as a later rollout phase.
Sources
- ABC Life Literacy Canada, "AI and the new digital divide: Why digital literacy matters more than ever," October 2025. abclifeliteracy.ca
- Statistics Canada, "Analysis on artificial intelligence use by businesses in Canada," second quarter of 2024 and 2026 releases. statcan.gc.ca
- Computerworld, "Global AI adoption is growing, and so is the digital divide," citing Microsoft AI Economy Institute, Jan 2026. computerworld.com
- OECD, "AI use by individuals surges across the OECD as adoption by firms continues to expand," Jan 2026. oecd.org
- Angus Reid Institute, "AI Divide: Canadians worry technology will widen inequality," survey of 1,842 Canadian adults, May 2026. angusreid.org
- CIRA, "Canada's Internet Factbook 2024." cira.ca
- Innovation, Science and Economic Development Canada, "Canada's National Artificial Intelligence Strategy: AI for All," 2026. ised-isde.canada.ca
- Prime Minister of Canada, "Prime Minister Carney launches AI for All," June 2026. pm.gc.ca
- Broadband Breakfast / Lincolnmain, citing CRTC 2024 Communications Monitoring Report on urban/rural 50/10 Mbps coverage. broadbandbreakfast.com
- Wikipedia, "Digital divide in Canada," overview of geographic and income-based access disparities. en.wikipedia.org
- Forbes, "Politics And The Perils Of AI: Exacerbating Social Divides In Canada," Feb 2025. forbes.com
- Holmes Consultants, "AI Adoption Statistics Canada 2026," compiling Statistics Canada series data. holmesconsultants.com