Quebec Survey Finds Workplace AI Benefits Uneven

Across the survey, 46% said AI reduced their workload, 37% saw no change, and 17% said it increased their workload. Stress showed a similarly mixed pattern: 28% reported less stress with AI, 27% reported more, and 44% reported no change.

Those results make the idea of universally "augmented" work too broad. The same tool can shorten one task while adding verification, rework, monitoring, or additional assignments elsewhere in the job. The study reports perceptions rather than independently measured productivity, so these percentages describe worker experience, not causal performance effects.

The report found a pronounced education gap. Fifty-five percent of respondents with postgraduate education said AI improved their productivity, compared with 22% of respondents whose highest credential was secondary school. Professionals also reported more autonomy and participation in AI decisions, while technical, service, manual, and industrial workers more often described imposed use, algorithmic monitoring, or fewer benefits.

Governance was limited across the sample. Only 12% said employees had been consulted before AI implementation, 26% considered their organization transparent about current and future AI use, and 35% expressed confidence in their employer's AI initiatives.

The survey was distributed by 11 union partners, participation was voluntary, and the 4,595 eligible responses form a non-probability sample. Public and parapublic workers were heavily represented, and response counts varied by question. The findings therefore should not be treated as a representative estimate for every Quebec worker or as proof that AI caused the reported outcomes.

For data and AI teams, the practical signal is that adoption rates alone are an incomplete success measure. Workload, stress, output quality, autonomy, and perceived surveillance can move differently across job groups. The report's own recommendations emphasize training, employee participation, clear organizational rules, and dialogue with unions; those controls matter because the benefits of workplace AI depend on how the system is introduced and governed, not merely whether it is available.