Official business surveys record what AI changed in staffing, tasks, and workflows, but none of those cited here measures dollars saved. In spring 2025, Statistics Canada found that 89.4 percent of Canadian businesses using AI reported no change in employment, and a New York Fed survey that August found that only 1 percent of service firms had laid off workers because of AI. The closest measure of savings is self-reported time: U.S. workers who used generative AI said it saved 5.4 percent of their work hours in a November 2024 survey.
Key takeaways
- Statistics Canada's spring 2025 survey reports effects on jobs, tasks, and workflows, and it does not report cost reductions.
- Among Canadian AI users, 6.3 percent reported lower employment and 4.3 percent reported higher employment because of AI.
- In the New York Fed's August 2025 survey, about 12 percent of AI-using service firms said they had hired fewer workers, while 11 percent of service firms said they had hired more.
- Self-reported time savings are the most direct public proxy for cost savings, and the St. Louis Fed's authors warn that saved time may not become output.
- A budget case for AI needs its own baseline, because no survey average transfers to a single workplace.
What Canadian businesses reported in spring 2025
Statistics Canada's analysis of AI use by businesses in the second quarter of 2025 draws on the Canadian Survey on Business Conditions, fielded from April 1 to May 5, 2025, and released on June 16, 2025. Its employment question went to the 12.2 percent of businesses that had used AI in the previous 12 months. Of those, 89.4 percent reported no change in employment, 6.3 percent reported a decrease, and 4.3 percent reported an increase. A year earlier, 8.8 percent had reported an increase, so the share citing job growth fell by about half.
The same survey asked how far AI had taken over tasks that employees used to do. Large extent: 5.3 percent. Moderate: 32.4 percent. Small: 47.2 percent. None: 15.1 percent. About 38 percent of AI users therefore saw a sizable shift of tasks, and most of the rest saw a small one. The release says nothing about what any of this saved or cost.
What changed inside the firm
Staffing was the smaller part of the story. The same release lists changes that AI-using businesses made to their operations, shown in the table with 2024 values for comparison. No dollar values are attached to any row.
| Change reported by Canadian AI users | Q2 2025 | Q2 2024 |
|---|---|---|
| Developed new workflows | 40.1 percent | 35.2 percent |
| Trained current staff to use AI | 38.9 percent | 38.5 percent |
| Purchased cloud services or storage | 25.7 percent | 16.1 percent |
| Worked with vendors or consultants | 18.2 percent | 11.3 percent |
| Hired staff already trained in AI | 4.9 percent | 8.2 percent |
| Made no changes | 22.1 percent | 28.1 percent |
The pattern is that adopting AI mostly meant reworking processes and teaching existing employees, and hiring specialists became less common. For a manager building a cost case, those rows are cost lines, not savings.
The U.S. view from the New York Fed
The Federal Reserve Bank of New York published results from its August 2025 regional business surveys on September 4, 2025. Among service firms, 40 percent used AI, up from 25 percent in 2024. Among manufacturers the figure was 26 percent, up from 16 percent. Only 1 percent of service firms reported layoffs due to AI in the past six months, against 10 percent a year earlier, and no manufacturer reported any in either year.
Hiring showed both effects at once. About 12 percent of AI-using service firms said they had hired fewer workers because of AI, while 11 percent of service firms and 7 percent of manufacturers said they had hired more. The authors report that the reduction in hiring was concentrated in jobs requiring a college degree. Just over a third of service firms and 14 percent of manufacturers that used AI said they had retrained workers. The survey covers the New York and northern New Jersey region only, so it does not stand in for the whole country.
Time saved: the nearest thing to a savings figure
A St. Louis Fed post dated February 27, 2025, by Alexander Bick, Adam Blandin, and David Deming, reported results from the Real-Time Population Survey. Workers who used generative AI said it saved 5.4 percent of their work hours in November 2024, about 2.2 hours in a 40-hour week. Counting nonusers, the saving was 1.4 percent of all hours. The authors estimated a 1.1 percent rise in aggregate productivity.
They also list limits. The time figures are self-reported, workers may spend saved time on leisure instead of output, and the gains say nothing about quality. The authors later revised some 2024 results after a change in question order, so the figures here are those of the February 2025 post. A saved hour is also not a saved dollar unless hours are cut or reassigned to paid work.
Checklist for a workplace savings estimate
- Which task is being changed, and what does it cost now in staff time?
- Is the saving measured by timing real work before and after, or taken from user reports?
- What do training, workflow redesign, and cloud services add on the cost side?
- Who reviews the output, and how many minutes does that take?
- Will saved time reduce overtime or contractor spending, or move into other work?
- What happens to the estimate if the tool is slower on difficult cases?
FAQ
Does AI reduce staffing costs in Canadian businesses?
Not in most reported cases. In the spring 2025 Statistics Canada survey, 89.4 percent of AI-using businesses reported no change in employment, and 6.3 percent reported a decrease.
Do any of these surveys measure money saved?
No. The Statistics Canada release and the New York Fed post report changes in jobs, hiring, training, and workflows. The St. Louis Fed work reports time saved, which is a proxy.
Why do the Canadian and U.S. figures differ?
The surveys ask different questions of different groups. Statistics Canada covers businesses across the country over 12 months, while the New York Fed covers firms in one region over six months, so the numbers should not be merged.
Is an unexpected benefit documented anywhere in these sources?
One candidate is that some firms added staff because of AI: 11 percent of service firms in the New York Fed survey said so. The surveys do not say which tasks those hires covered.
Related reading
More in AI Tools and Industry.
Last verified: October 2026
Sources
- Analysis on artificial intelligence use by businesses in Canada, second quarter of 2025, Statistics Canada, June 16, 2025
- Are Businesses Scaling Back Hiring Due to AI?, Liberty Street Economics, Federal Reserve Bank of New York, September 4, 2025
- The Impact of Generative AI on Work Productivity, Bick, Blandin, and Deming, Federal Reserve Bank of St. Louis, February 27, 2025