AI continues to be blamed for a spate of job cuts, with Cisco (CSCO) the latest tech firm to jump on the trend this week. And with this trend coming alongside an overall slowdown in hiring in the US, AI’s role in the labor story has become primarily seen as negative.
Squaring this trend with the available data on how companies have — and haven’t — changed their workforces in the aggregate, however, paints a more complicated picture of both the US labor market and AI’s role in reshaping it.
Job openings may be lower in certain sectors deemed vulnerable to AI, but a new analysis from researchers at the New York Fed suggests that may not be due to the technology itself.
The New York Fed’s paper scrutinized vacancies in fields considered vulnerable to automation, based on a measure developed by economists at Anthropic that broke down occupations where many tasks are both doable by AI and already done by AI in a work setting, utilizing usage data.
The most AI-exposed occupations, by that metric, are computer programmers, customer service representatives, and data entry keyers. The New York Fed researchers examined whether occupations with high or low AI exposure saw a significant difference in hiring pre- and post-ChatGPT in late 2022.
If AI had a noticeable impact, the researchers wrote, hiring patterns would have moved similarly across fields with high and low AI exposure before ChatGPT’s release and diverged — and kept diverging — after. What they found was that while there had been “a relative decline in postings for occupations with higher AI exposure,” the trend had pre-dated ChatGPT.
“The divergence between high- and low-exposure occupations began before 2022 and does not show a clear additional break in trajectory after 2022,” the researchers wrote. “Besides, the gap in labor demand between high- and low-exposure jobs stabilizes after 2023, at odds with AI gradually displacing exposed occupations.”
The AI narrative is similarly messy when looking at official government data: Hiring rates began to head lower in early 2022, but picked up in March to reach their best level in two years. And while layoff rates have ticked up more recently, all while tech companies cite AI as a rationale for job cuts, they remain relatively low and have been hovering between 0.9% and 1.2% since 2021.
Drilling deeper to compare the difference in senior and junior-level job postings in occupations with high AI exposure post-ChatGPT, the New York Fed researchers also found “the slowdown in postings is not concentrated specifically in entry-level highly exposed jobs,” countering another narrative that says AI shocks are largely responsible for young people’s difficulties in getting hired.
“Overall hiring has slowed since 2022, and unemployment has increased among young workers and recent college graduates,” the New York Fed researchers said. “The evidence from job postings suggests that while AI may be contributing to recent labor market developments, it is not the main driver of the slowdown in hiring.”
Meanwhile, a separate analysis published this week by Michael Pearce, chief US economist at Oxford Economics, showed that while AI adoption has become mainstream in leading sectors, “usage still appears relatively low, explaining the mostly muted impacts on aggregate productivity and the labor market so far.”
The unemployment rate for AI-exposed occupations has actually dropped, along with the overall unemployment rate, since December, “consistent with signs that broader labor-market conditions have improved,” Pearce wrote.
One possible bellwether for AI’s eventual impact on labor, however, might be the information sector, where AI adoption is high and there’s been a jump in hires and fires as the “net change in jobs — hires minus layoffs — remains little changed.” Labor churn in that corner of the job market has surged.
If that trend were to bleed into multiple sectors all at once, it could drive unemployment higher as more displaced workers face a market that demands a different set of skills. But “the baseline forecast assumes that AI proves more labor-augmenting than labor-displacing.”
Goldman Sachs Research economist Elsie Peng argued in a report this week that even though jobs with high exposure to AI substitution have seen openings slip below pre-pandemic levels — as those with less exposure have fallen more gradually — labor market mismatch has actually declined, perhaps quelling some concerns “that the type of workers the economy needs is changing faster than workers.” Several occupations clocking high exposure levels started out with big labor shortages, they noted, which helps.
“While the first stage of AI deployment has been fortuitously timed because it coincided with a labor shortage in the most AI-exposed occupations, the next stage of deployment will likely require more adaptation by the workforce,” Peng wrote.
