ChatGPT didn’t trigger the hiring slowdown, NY Fed study Says

NEW YORK, UNITED STATES — The decline in job postings for AI-exposed roles began before ChatGPT’s December 2022 release and shows no clear acceleration after it, Federal Reserve Bank of New York economists found in a study analyzing hiring data through January 2026.

Hiring slowdown in AI-exposed roles predates ChatGPT

The researchers tested whether positions requiring AI-adjacent skills declined after ChatGPT’s release, using occupation-level AI exposure metrics developed by Anthropic and job posting data from Lightcast. 

“The divergence between high- and low-exposure occupations began before 2022,” the study found, with no clear break in trend at the moment generative AI went mainstream.

Less than 10% of workers and vacancies are in occupations with an AI exposure score of at least 0.4 on a scale of zero to one; 40% of workers are in jobs with zero measured AI exposure. The study, by New York Fed economists Richard Audoly, Miles Guerin, and Giorgio Topa, covers hiring data through 2024 and job postings through January 2026.

No AI effect found on entry-level hiring

The researchers also tested whether AI was disproportionately eliminating entry-level roles — a central fear in current workforce debates — and found no support in the data. 

“No clear divergence in labor demand between junior and senior positions” was observed in AI-exposed occupations, the study noted, adding that declines in those postings track the broader softening in hiring across all occupation types.

Overall hiring has slowed since 2022, but the NY Fed data shows that slowdown predates ChatGPT and is not concentrated in AI-exposed occupations.

New York Fed business surveys found firms plan to incorporate AI primarily through worker retraining, not headcount reduction.

The study concludes the evidence “provides little indication of a distinct AI-driven decline in labor demand,” making the current hiring slowdown a broader macroeconomic story, not an AI story.

For BPO providers, the NY Fed findings cut against the most alarming near-term forecasts. The data shows AI’s labor market footprint — at least through early 2026 — is concentrated in a narrow slice of highly exposed occupations, not the frontline roles that drive offshore staffing demand. 

Providers who frame their AI positioning around retraining rather than replacement are aligning with the evidence — and are better placed to extend  client relationships rather than defend against disruption narratives.