Survey finds income and education shape who benefits most from AI tools
Canadians are turning to artificial intelligence more often than ever but rising usage has done little to ease widespread doubts about who actually stands to gain from the technology.
A majority now believe AI is more likely to widen inequality in society than narrow it, and most reject the idea that everyone has an equal shot at benefiting from it, regardless of where they live or who they are.
That’s according to new polling from the Angus Reid Institute, based on a survey of 1,842 Canadian adults which found that 56% of respondents said AI is more likely to create inequality, compared with just 7% who think it will make things more equal.
A larger share (68%) disagreed that everyone has the same opportunity to benefit from the technology.
That skepticism holds steady across every income bracket. Even among households earning $200,000 or more, 62% disagreed that AI opportunity is evenly distributed, while the figure climbed to 73% among those earning under $50,000.
Usage climbs despite the doubts
Daily engagement with AI tools has grown substantially since late last year. The share of Canadians reporting they use AI several times a day rose from 10% in November to 16% by June. Usage also ticked up among those turning to AI once a day, and several times a week, over the same period.
Yet only 16% of respondents landed in the "positive" category on the Institute's AI Assessment Index, a measure built from four questions gauging trust and perceived impact. Nearly two-thirds fell into the negative (43%) or critical (24%) categories, with the remaining 17% uncertain.
Familiarity with the technology did little to shift those views. Majorities across every assessment group, from the most positive to the most critical, said they know at least a little about AI and its applications, suggesting the unease isn't rooted in a lack of exposure.
Income divide in how people engage
Perhaps the clearest fault line in the data is between active and passive AI use.
Canadians with university degrees and higher incomes were far more likely to say they actively prompt AI tools directly, while those with less formal education or lower earnings were more likely to describe their exposure as incidental, arriving through built-in features in apps and platforms without being sought out.
Active engagement peaked among households earning $150,000 or more, with 46% to 48% in that bracket saying most of their AI use is deliberate. That compares with just 21% among households earning under $50,000. A similar pattern emerged by education level: 41% of university-educated respondents described their use as active, versus 23% of those with a high school education or less.
The gap matters because passive exposure correlates strongly with distrust. Two-thirds of Canadians in the "critical" assessment group said most of their AI exposure comes unsolicited, compared with just 17% of those with a positive outlook, who were far more likely to be hands-on users.
Where Canadians see value, and where they don't
Asked what benefits AI offers, 43% pointed to faster learning and research, while 37% cited easier access to information and 35% said it saves time.
Fewer respondents saw AI delivering more targeted gains, such as personalized recommendations (17%) or help with work or school (13%). Notably, 19% said they see no benefit at all, a figure that jumped to 33% among those with no exposure to large language model tools such as ChatGPT, Claude, or Perplexity.
On the concern side, the top worry, cited by 41% of respondents, was that reliance on AI will erode critical thinking and problem-solving skills.
Difficulty distinguishing real from AI-generated content followed at 36%, while 29% flagged the spread of misinformation. Smaller but still notable shares cited inaccurate outputs, job displacement, privacy, and overdependence on the technology, each in the mid-to-high 20% range.
Trust in AI-sourced information remains thin. Just 1% of respondents called it very trustworthy, while 43% said it was trustworthy enough. Nearly seven in 10 Canadians disagreed that AI-generated misinformation is a problem that will eventually be resolved.
