Marketing professionals rank among the most vulnerable to AI disruption, with Indeed recently placing marketing fourth for AI exposure.
But employment data tells a different story.
New research from Yale University’s Budget Lab finds “the broader labor market has not experienced a discernible disruption since ChatGPT’s release 33 months ago,” undercutting fears of economy-wide job losses.
The gap between predicted risk and actual impact suggests “exposure” scores may not predict job displacement.
Yale notes the two measures it analyzes, OpenAI’s exposure metric and Anthropic’s usage, capture different things and correlate only weakly in practice.
Exposure Scores Don’t Match Reality
Yale researchers examined how the occupational mix changed since November 2022, comparing it to past tech shifts like computers and the early internet.
The occupational mix measures the distribution of workers across different jobs. It changes when workers switch careers, lose jobs, or enter new fields.
Jobs are changing only about one percentage point faster than during early internet adoption, according to the research:
“The recent changes appear to be on a path only about 1 percentage point higher than it was at the turn of the 21st century with the adoption of the internet.”
Sectors with high AI exposure, including Information, Financial Activities, and Professional and Business Services, show larger shifts, but “the data again suggests that the trends within these industries started before the release of ChatGPT.”
Theory vs. Practice: The Usage Gap
The research compares OpenAI’s theoretical “exposure” data with Anthropic’s real usage from Claude and finds limited alignment.
Actual usage is concentrated: “It is clear that the usage is heavily dominated by workers in Computer and Mathematical occupations,” with Arts/Design/Media also overrepresented. This illustrates why exposure scores don’t map neatly to adoption.
Employment Data Shows Stability
The team tracked unemployed workers by duration to look for signs of AI displacement. They didn’t find them.
Unemployed workers, regardless of duration, “were in occupations where about 25 to 35 percent of tasks, on average, could be performed by generative AI,” with “no clear upward trend.”
Similarly, when looking at occupation-level AI “automation/augmentation” usage, the authors summarize that these measures “show no sign of being related to changes in employment or unemployment.”
Historical Disruption Timeline
Past disruptions took years, not months. As Yale puts it:
“Historically, widespread technological disruption in workplaces tends to occur over decades, rather than months or years. Computers didn’t become commonplace in offices until nearly a decade after their release to the public, and it took even longer for them to transform office workflows.”
The researchers also stress their work is not predictive and will be updated monthly:
“Our analysis is not predictive of the future. We plan to continue monitoring these trends monthly to assess how AI’s job impacts might change.”
What This Means
A measured approach beats panic. Both Indeed and Yale emphasize that realized outcomes depend on adoption, workflow design, and reskilling, not raw exposure alone.
Early-career effects are worth watching: Yale notes “nascent evidence” of possible impacts for early-career workers, but cautions that data are limited and conclusions are premature.
Looking Ahead
Organizations should integrate AI deliberately rather than restructure reactively.
Until comprehensive, cross-platform usage data are available, employment trends remain the most reliable indicator. So far, they point to stability over transformation.
