Is AI Replacing Jobs? What the New York Fed's Survey Actually Found
The New York Fed found 61% of service firms now use AI but only 4% laid anyone off because of it. The bigger shift is quieter: roles that are simply never posted.
Quick answer
Mostly not through layoffs, so far. The New York Fed's August 2026 surveys found 61% of service firms use AI, but only 4% laid off workers because of it. About 15% hired fewer people than they otherwise would have, and 13% hired more. The main effect is roles that are never posted, not people being fired.
Is AI replacing jobs? According to one of the more careful data sources available, mostly not through layoffs, at least not yet. But the same data shows a quieter change that matters just as much for anyone running or selling to a service business.
What the New York Fed measured
On 1 September 2026, economists at the Federal Reserve Bank of New York published results from their regional business surveys, conducted in August across New York and northern New Jersey. They compared the answers with the same questions asked in 2024 and 2025.
| Share of firms using AI | 2024 | 2025 | 2026 |
|---|---|---|---|
| Service firms | 25% | 40% | 61% |
| Manufacturers | 16% | 26% | 51% |
Adoption has more than doubled in two years for service firms and roughly tripled for manufacturers.
Are companies laying people off because of AI?
Very few, in this survey. Only 4% of service firms reported laying off workers because of AI, up from 1% the year before. No manufacturers reported AI-related layoffs at all.
That is a useful counterweight to headlines that treat every workforce reduction as an AI story. We covered the same labeling problem in the Challenger August report, where AI dropped to the fourth most-cited reason for cuts.
The number that matters more: hiring
The survey also asked whether AI changed hiring. About 15% of service firms said they hired fewer workers than they would have without AI. About 13% said they hired more because of it.
Put next to the 4% layoff figure, that changes the picture. The most common way AI is reducing labor is not by firing people. It is by making the next hire unnecessary.
- A layoff is visible. It gets announced, reported and counted.
- An unposted role is invisible. No tracker records a job that was never advertised.
- In this survey, firms hiring fewer outnumber firms laying off by almost four to one.
Adoption is wide but shallow
The headline adoption rate can also mislead. Although 61% of service firms use AI, the median share of their workers using it is just 17%. For manufacturers it is 7%. Three quarters of service firms, and more than 90% of manufacturers, describe their AI investment as minimal to modest.
Just over a third of service firms reported retraining workers. The authors' conclusion is that firms are "investing in their existing workforces rather than replacing large swaths of people."
In practice, most businesses have bought or enabled AI tools without yet redesigning the work around them. That suggests the bigger effects are still ahead, not behind.
What this means for service businesses
For agencies, consultancies and other firms that sell expertise, the pattern matters in two directions.
First, inside your own business, AI pressure is more likely to show up as roles you do not refill than as people you let go. That is easier on everyone in the short term, and it quietly changes what juniors can learn and who carries the load.
Second, your clients are going through the same process. They are unlikely to fire their teams. They are likely to not replace the next person who leaves, and then to look at external spend with the same question in mind.
Why the low layoff number is not reassuring
A low layoff rate can read as good news. It partly is. But layoffs are the loud, contestable version of workforce change. They get debated, sometimes reversed, and they show up in data.
Hiring that never happens is quiet and usually permanent. If that is already occurring at several times the layoff rate while adoption is still shallow, the long-run effect on demand for routine professional work could be larger than the layoff figures suggest.
What to do with this
- Track unfilled roles, not just headcount. Note which roles went unfilled this year and where that work went. That is your real AI adoption record.
- Make redistributed work deliberate. If a role is not being refilled, automate the repetitive part properly instead of letting it land on whoever has capacity.
- Ask clients about their hiring plans. A client that is not backfilling roles will soon be reviewing agency scope. Hearing it early gives you time to shape the answer.
The New York Fed data is reassuring about mass layoffs and sobering about everything else. The change is real. It is just happening in job ads that never go live.
Frequently asked questions
- How many businesses use AI in 2026?
- In the New York Fed's regional surveys, 61% of service firms reported using AI in 2026, up from 40% in 2025 and 25% in 2024. Among manufacturers, 51% reported using AI, up from 26% and 16%.
- Are companies laying off workers because of AI?
- Some are, but few in this survey. 4% of service firms reported laying off workers due to AI in 2026, up from 1% the year before. No manufacturers reported AI-related layoffs.
- If layoffs are rare, how is AI affecting jobs?
- Mainly through hiring. About 15% of service firms said they hired fewer workers than they would have without AI, while about 13% said they hired more. The effect shows up as positions not created rather than people let go.
- How deeply are businesses using AI?
- Not very deeply yet. The median share of workers using AI was 17% at service firms and 7% at manufacturers, and three quarters of service firms described their AI investment as minimal to modest.