The widespread replacement of workers by AI won’t happen until a recession strikes, according to IMF first deputy managing director Gita Gopinath.
Since OpenAI’s ChatGPT kicked off the generative AI revolution in late 2022, there has been extensive handwringing that the technology would result in mass layoffs.
The promise of a tool that could understand complex questions posed in normal, everyday speech and spit out cohesive answers seemed poised to end a host of professions ranging from customer service reps to journalists.
Ultimately, those concerns haven’t panned out as the most fearful had imagined, largely because the economy was on solid footing. However, if things should take a turn for the worse, then the labor market could start to experience an “abrupt shift” of human workers being replaced by AI.
“You see it when the recession strikes,” Gopinath said during an onstage interview at the Coins2Day Global Forum in New York. “That’s when you really see the effect of automation on the labor market.”
When the economy is strong companies don’t need to cut their headcount and prefer to be fully staffed, Gopinath argues.
“When you have good times, and companies are making lots of profits, they hold on to workers,” she said. “They’re investing in the technology [AI], but they’re holding on to workers.”
The corporate belt-tightening of a recession forces companies to lay off workers.
However, once the economy emerges from that recession some of those jobs never return.
This phenomenon, which economists call a jobless recovery, happened after the Great Recession permanently gutted American manufacturing jobs.
A “jobless recovery is a reflection of companies finally letting go of workers, fully automating the task, and not hiring both workers back,” Gopinath said.
The extent to which AI will impact the labor market is a hotly debated topic.
A widely cited 2023 report from Goldman Sachs estimated that AI could impact roughly 300 million jobs in the U.S. And Europe. Not all those jobs would see the entirety of their duties automated away, with some only seeing between 25% and 50% of their workloads taken over by AI.
“So the caution here for policymakers is, just because you’re not seeing it now doesn’t mean you won’t see an abrupt shift in a recession,” Gopinath said.
That said, AI, like all technologies, is not merely a deadweight on the global economy.
Technologies carry with them the promise of productivity gains, where the economy can produce more goods and services with fewer resources.
An April research paper by Nobel laureate Daron Acemoglu projected that AI could lead to 0.7% increase in productivity over 10 years in the U.S.
Acemoglu termed that amount “nontrivial but modest.”
During her onstage interview, Gopinath cited the IMF’s slightly more bullish range of estimates for global productivity growth of between 10 to 80 basis points of growth over five years as a result of AI.
“If you’re looking at a baseline of growth of 3% if you have an 80 basis points increase that’s a pretty strong effect,” she said.
The latest IMF projections expect global GDP growth of 3.1% over the next five years. According to Gopinath, those productivity increases would be critical to the well-being of the global economy in the long term because they would boost that positive but sluggish growth rate.
“Growth, while holding up, is not a good number,” she said. “We would like that growth to be higher and one of the main factors that can raise it is productivity growth…so in that sense, this new technology sounds exciting.”