The AI job cuts are here - or are they?

TL;DR: Amazon’s announcement of 14,000 corporate job cuts fed into anxiety about AI replacing workers, but experts question whether AI is fully to blame. Yale University researcher Martha Gimbel calls extrapolating from executives’ remarks during cuts “possibly the worst way” to determine AI’s effects on jobs, noting company-specific dynamics and economic cycles may be driving layoffs.

Amazon’s move this week to slash thousands of corporate jobs fed into a longstanding anxiety: that Artificial Intelligence is starting to replace workers. The tech giant joined a growing list of companies in the US that have pointed to AI technology as a reason behind layoffs.

Context and Background

Chegg, the online education firm, cited the “new realities” of AI as it announced a 45% reduction in workforce on Monday. When Salesforce cut 4,000 customer service roles last month, its chief executive said AI agents were doing the work. UPS said on Tuesday that it has cut 48,000 jobs since last year. The delivery company’s chief executive previously linked redundancies, in part, to machine learning.

But extrapolating from executives’ remarks during cuts is “possibly the worst way” to determine the effects of AI on jobs, said Martha Gimbel, executive director of the Budget Lab at Yale University. Company-specific dynamics, she said, are often at play.

“There is a real tendency, because everyone is so freaked out about the possible impact of AI on the labour market moving forward, to overreact to individual company announcements,” Ms Gimbel said.

Evidence: Office Workers Most Affected

Certain subsets of the workforce—recent college graduates and data centre employees, for example—in fact are particularly vulnerable to the technology’s adoption. A recent study from the Federal Reserve Bank of St Louis found a correlation between occupations with a higher prevalence of AI and increases in unemployment since 2022.

But Morgan Frank, assistant professor at the University of Pittsburgh, has studied unemployment risk by occupation and found that the only workers affected by the launch of ChatGPT in November 2022 were in the office and administrative support sector.

For them, their probability of claiming unemployment jumped in early 2023, he said—immediately following the entrance of the chat bot developed by OpenAI. But for computer and maths occupations, “there is no discernible change in the trend around the launch of ChatGPT,” he said.

“Both tech workers and admin workers—they’re in a rougher job market than they were in a couple years ago,” Mr Frank said. “I’d be sceptical that AI is the reason for all of it, though,” he added.

Typical Patterns of Hiring and Firing

Amazon and many of its rivals in the tech sector hired at a rapid clip in the years leading up to the coronavirus pandemic and in the pandemic’s early months, when the Federal Reserve lowered US interest rates to near zero. The hiring set these firms up for eventual workforce reductions, experts said—a dynamic separate from the generative AI boom over the last three years.

The Fed also started hiking interest rates around the time that ChatGPT was launched.

“A lot of this conversation feels very different to people because the phrase AI is in it,” said Ms Gimbel, of the Budget Lab. “But so far, nothing that I’ve seen looks different than typical patterns of companies hiring and firing, particularly at this point in an economic cycle.”

A big question, she added, is what hiring patterns look like when the economy returns to a period of solid growth.

In the long run, Ms Gimbel said, breaking out the cyclical versus AI-driven job losses will be crucial. If, for instance, the US economy were to fall into a recession, human resources and marketing jobs would be the expected casualties. Those jobs, though, also happen to be exposed to AI, complicating the task of identifying whether the cuts are a result of macroeconomic conditions or the technology’s adoption—or both.

Amazon at the Forefront

Amazon, which confirmed it plans to cut roughly 14,000 corporate roles, said it needs to be “organised more leanly” to seize the opportunity provided by AI. The company has been performing well. It reported quarterly results in July that beat Wall Street expectations on several counts, including a 13% year-over-year increase in sales to $167.7bn (£125bn).

Enrico Moretti, an economics professor at the University of California, Berkeley, said that the largest tech companies like Amazon are at the forefront of AI-related job cuts, “in part because they’re both producers and consumers of AI”.

Still, he acknowledged that a correction following robust hiring during the pandemic also may have driven the company’s latest round of layoffs.

Amazon is likely able to automate jobs faster than most of its rivals because of its scale, said Lawrence Schmidt, an associate professor of finance at the MIT Sloan School of Management.

“It’s not at all crazy to think that Amazon might want to shed certain types of roles, or refrain from hiring additional people in certain types of roles, if they can be quickly automated,” Mr Schmidt said. “Regardless of what happens to counts of jobs overall,” he added, “you would expect there to be reallocation.”

Looking Forward

The challenge for researchers and policymakers lies in disentangling genuine AI-driven displacement from cyclical economic patterns. The timing coincidence—Federal Reserve rate hikes beginning around ChatGPT’s launch—complicates attribution of cause.

For organisations evaluating AI implementation, Amazon’s case suggests scale matters significantly in automation capability. Smaller organisations may face longer timelines for achieving similar automation effects, potentially providing more runway for workforce adaptation.

The evidence from Morgan Frank’s research—that office and administrative support sectors showed immediate unemployment increases post-ChatGPT whilst technical roles did not—suggests AI’s employment impact may be more nuanced and role-specific than broad “AI will replace jobs” narratives suggest.

What remains clear is that attributing specific layoff decisions to AI based solely on executive statements during redundancy announcements provides insufficient evidence for understanding the technology’s actual employment impact.

Source Attribution:

Share this article