AI Hasn’t Taken a Lot of Jobs Yet, Despite Apocalyptic Predictions: Yale Study

Three years after ChatGpt exploded into the scene, American workers are still showing up in their work. A new study from Yale University’s Budget Institute and the Brookings Institute looked into federal employment data until July, finding that AI has not yet caused the mass unemployment that technology executives are predicting.

Researchers tracked how quickly the mix of occupations has changed since November 2022, when Openai opened ChatGpt to the public. Although employment has shifted slightly faster than in recent years, one percentage point higher than the internet boom of the early 2000s, due to the technological transition, rather than the economic upheaval that many fear.

Winter may still be here, but it’s not now.

“We’re not a whole economy’s work apocalypse right now. It’s pretty much stable,” said Molly Kinder, a senior Brookings fellow and co-author of the paper. Financial Times. “It should be a encouraging message to the anxious public.”

The gap between Silicon Valley rhetoric and the reality of workplaces has become severe. As reported by DecryptionWhile humanity’s CEO, Dario Amodo said up to 50% of the entry-level white-collar roles could disappear within five years, Jeffrey Hinton, known as the “AI Godfather,” estimated that AI could dramatically exacerbate the financial gap if things were going on.

“What really happens is that the rich use AI to replace workers,” he said. Financial Times. “It will generate massive unemployment and a significant rise in profits. It will enrich the few people much and make most people poorer. It’s not the AI’s fault, it’s the capitalist system.”

In a recent study claiming that AI is consistent with skilled human workers in at least 44 business areas, Sam Altman has repeatedly chosen customer service jobs as particularly vulnerable.

But the data tells a different story. Yale researchers have examined several metrics, including changes in occupational mix, industry-specific shifts, and levels of AI exposure across a variety of occupations. In theory, based on Openai’s own metrics, workers in occupations most exposed to AI automation showed no signs of displacement. Approximately 18% of workers hold employment in the highest AI exposure category, a percentage that has remained flat since January 2023.

The information sector, including newspapers, film and data processing, showed the biggest occupational change. However, these changes began before the release of ChatGpt and suggest industry-specific factors rather than AI disruption. Finance and professional services showed similar patterns. This precedes the AI ​​revolution.

Young university graduates are struggling, with the unemployment rate for ages 20-24 rising from 4.4% in April to 9.3% in August’s bachelor’s degree. However, the research team found that it coincides with the pattern of older degree holders ages 25-29, indicating a broader labor market slowdown rather than an alternative to AI. Dissimilarities between these age groups have fluctuated within a narrow range of 30-33% since 2021, with no acceleration after ChatGPT arrival.

Historical precedents support researchers’ skepticism about immediate confusion. Computers did not become standard office equipment until almost ten years after their public release. The transformation of the internet workplace has grown even longer. The study notes that occupational changes peaked during the period of large industrial shifts at a rate of 20-21% between the 1940s and 1950s. The current change is about 10%.

“Historically, widespread technical disruption in the workplace tends to occur over decades rather than months or years,” the researchers write.

The researchers acknowledged the limitations of important data. Openai’s “exposure” metric measures theoretical vulnerabilities rather than actual AI use. Human usage data from Claude Chatbot does not represent a broader workforce pattern, indicating a great focus between coders and writers. The team sought comprehensive usage data from all major AI companies to properly assess workplace impact.

The research team plans monthly updates to track new patterns. For now, almost three years after the AI ​​revolution, the most dramatic workplace change isn’t what executives actually do to hire at scale, but how many executives talk about AI.