For a long time, the public has generally believed that the impact of AI on the job market began with the release of ChatGPT in November 2022. However, a new study led by Morgan Frank from the University of Pittsburgh, through big data analysis, has challenged this long-held belief. The research team thoroughly analyzed U.S. Department of Labor unemployment data, 10.6 million LinkedIn profiles, and 3 million university courses, discovering that the employment prospects for AI-related jobs had already started to decline as early as early 2022—several months before the surge in large language model popularity.

ChatGPT

The study shows that in the spring of 2022, occupations heavily affected by large language models, such as computer and mathematical fields, had already begun to experience rising unemployment risks. Although during the pandemic, these technical professions showed strong employment resilience due to their ability to work remotely, their unemployment risk was once 20% to 80% lower than traditional sectors like construction. However, this advantage quickly narrowed in early 2022.

The study questions the simple narrative that "ChatGPT is the starting point of the unemployment wave," indicating that the labor market's response to technological change is much earlier and more sensitive than the public perceives. Companies may have already been adjusting their talent structures in anticipation of automation trends before the widespread adoption of generative AI.