Executive Summary

A recent Empirical Intelligence survey (N=1,004 Canadian citizens) found that Canadians are divided over AI being used by banks to decide whether to approve people for credit cards and by doctors to assist with medical diagnoses. Small majorities are at least slightly comfortable with each AI application, while two out of five are not comfortable at all with either use case.

Key Findings

  • Gender and education are the biggest divides: most university and postsecondary graduates as well as men are comfortable with each use of AI, while less than half of Canadians with a high school education or less and women are comfortable with each AI use.
  • Gap between cities and rural areas: seven to nine percentage points more city residents than residents of small towns and rural areas are at least slightly comfortable with each AI use.
  • There's an age gap for AI use by banks but not by doctors: Canadians aged 55 and over are 12 points less comfortable than Canadians under 55 with AI use by banks for credit card decisions. Comfort with AI use by doctors for diagnoses is similar in both age groups.


In a recent pan-Canadian survey with 1,004 respondents (Canadian citizens 18+), Empirical Intelligence collaborated with Behavieural to examine public comfort with artificial intelligence in high-stakes contexts.

The questions asked respondents how comfortable they are with AI being used for two high-stakes decisions: a doctor using AI to assist with medical diagnoses and a bank deciding whether to approve them for a credit card.

About half of Canadians are at least slightly comfortable with AI being used for high-stakes decisions. Fifty-five percent are at least slightly comfortable with a bank using AI to decide whether to approve them for a credit card, while 53% are at least slightly comfortable with a doctor using AI to help diagnose their condition. Despite the very different contexts, Canadians express remarkably similar levels of comfort with AI in banking and medicine. But the public is far from enthusiastic: about two in five Canadians are not comfortable at all with either use.

Canadians are divided in their comfort with AI use for high-stakes decisions

Gender and education are the biggest divides in comfort with AI

Some groups are more comfortable with the two uses of AI than others. The following figures show the percentages of respondents in each demographic category who are at least slightly comfortable with each use of AI.

The biggest gaps are between gender and education groups. Men were 15 to 17 percentage points more comfortable with each AI use than women. University graduates were 10 to 19 percentage points more comfortable with each AI use than respondents with high school or less.

Who is comfortable with AI use?

There were also gaps associated with the type of community and age. Respondents in cities were seven to nine percentage points more comfortable with each use of AI than respondents in small towns or rural areas. Finally, respondents under 55 were 12 percentage points more comfortable than respondents 55 and over with banks using AI to make credit-card approval decisions. There was no age gap for doctors using AI to diagnose a condition.

Understanding differences in comfort with AI may be key to gaining public acceptance

Small majorities of Canadians are at least slightly comfortable with AI use by banks and doctors. However, companies developing AI systems for banks and doctors and professionals implementing them should not ignore the two out of five Canadians who are not comfortable at all with each use of AI. They should also try to understand the concerns of groups who are more reluctant to accept each use of AI.

Most importantly, women are less comfortable with each use of AI than men. Canadians without a university education are, similarly, less comfortable than Canadians with a university education. Residents of small towns and rural areas are also less comfortable than residents of cities and suburbs. Finally, Canadians 55 and over are less comfortable with AI use by banks for decisions about credit cards than Canadians under 55. There is no such gap for AI use by doctors for their diagnoses.

The results suggest that public acceptance of AI in high-stakes decisions is neither uniformly positive nor negative. Instead, comfort varies substantially across groups, particularly by gender and education. For companies and professionals deploying AI in these settings, understanding these differences may be crucial to gaining public acceptance.

Methodology

Sample: 1,004 Canadian citizens aged 18 and over. The survey was run online on August 21-23, 2026. Respondents were recruited from a variety of online panels using Cint Exchange. Results are weighted to Census benchmarks for age, gender, education, and region, and then weighted halfway toward 2025 federal election results using respondents' recalled vote. If the results were from a probability sample, the margin of error would be 3.1 percentage points, 19 times out of 20. Interviews were in English and French.