For years, dialogue of how AI would possibly change duties, the workplace, and even workers themselves has dominated the cultural zeitgeist. However till now, there hasn’t been a lot overarching information on how AI is definitely altering the workforce.
On August 26, researchers at Stanford printed what they’re calling the “largest scale, most real-time effort” to calculate that influence—and it consists of some dangerous information for Gen Zers.
The research was performed by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, all researchers at Stanford’s Institute for Human-Centered AI. They used an administrative dataset from ADP, the most important payroll software program supplier within the U.S., to trace month-to-month payroll data throughout tens of hundreds of companies from late 2022 to July 2025.
That information was supplemented by a calculation of “occupational AI publicity” in a various array of fields—primarily utilizing information from prior research (see right here and right here) to categorize how a lot numerous professions are being impacted by AI know-how.
Based mostly on this evaluation, the researchers reached six key conclusions on how AI is altering the workforce. Listed below are crucial findings for workers to grasp.
Unhealthy information for Gen Zers in AI-exposed fields
The research reveals that, for some early profession professionals, AI might certainly be making it more durable to discover a job—even if, total, employment “continues to develop robustly.”
Amongst employees aged 22 to 25, the researchers uncovered a 13% decline in employment in occupations most uncovered to AI, like software program builders and customer support representatives, even after controlling for firm-level impacts. For instance, by July 2025, employment for software program builders aged 22 to 25 declined by almost 20% in comparison with its peak in late 2022.
In an interview with Quick Firm, Brynjolfsson clarified that his staff can’t essentially draw a tough causal line between corporations’ AI publicity and this decline in employment. Nevertheless, he says, the sample did stay constant even when testing towards a spread of variables, which suggests a probable connection.
“We didn’t do an experiment the place we gave some corporations LLMs and never different corporations, so we’re simply observing what occurred,” Brynjolfsson says. “However we may rule out among the foremost various hypotheses. Even in the event you take out the complete tech business, as an example, you continue to see this phenomenon: It’s one thing occurring past simply tech. Or in the event you take away all distant work and simply have a look at jobs that aren’t distant, you see the identical factor.”
Excellent news for older employees
Whereas jobs for younger employees in AI-exposed fields have been declining, older employees have been largely shielded from these detrimental impacts. Employment for mid-level and senior workers in AI-exposed fields has really elevated by 6 to 9 % from 2022 to 2025, in keeping with the expansion throughout the job market at giant.
“I feel one of many issues that the senior folks might have had a bonus in is that they had lots of on-the-job studying ideas and methods about easy methods to use AI that aren’t essentially taught at school,” Brynjolfsson says. “That was one thing that gave them a differentiation from what the LLMs had been capable of do. However for all of them—mid-level, senior, junior—I feel you must keep watch over the quickly evolving capabilities. This isn’t the tip of AI advances.”
Excellent news for everybody in an occupation that doesn’t depend on AI (but)
In jobs which are much less uncovered to the impacts of AI tech—like nursing aides, upkeep employees, and taxi drivers—employment remained secure or continued to develop. In truth, younger employees in these sorts of professions really noticed larger employment progress than their older counterparts.
Not all AI use is created equal
In keeping with the researchers, not all makes use of of AI are related to declines in employment. Employment declined for younger employees in occupations that largely depend on AI to automate duties (i.e. full them of their entirety), whereas employment really grew in occupations the place AI was primarily augmentative (i.e. serving to the person full a job or be taught a brand new ability).
“The fact is that there are much more advantages in augmenting and permitting folks to do new issues,” Brynjolfsson says. “[Automating and augmenting] will be productive, however I feel complementing employees, slightly than substituting for employees, has been underrated. It was putting to see the information present the identical factor.”
