Generative artificial intelligence (AI) has ceased to be the future and has become the present. Millions of workers and companies have integrated into their daily activities tools capable of generating text, images, and video. One of the main objectives? Achieving productivity gains that translate into greater efficiency of production processes.
But the effects of AI on the labor market can go far beyond productivity. If these tools change the way we work, they can also alter our relationship with the labor market and the opportunities we encounter outside our current role.
In 2025, one in four American workers used generative AI in their workplace. Among those who used it, 81% stated that these tools helped them save time, and more people were optimistic that the technology would advance their professional careers than those who feared it might replace them. Such optimism raises a less explored question: Is generative AI also improving workers’ opportunities beyond their current position? Workers can use these tools to identify vacancies, compare alternatives, or prepare applications. However, looking more broadly does not necessarily mean finding something better. Generative AI can facilitate exploring other jobs without increasing the probability of leaving the company. Differentiating between greater job search and greater labor mobility is, therefore, key to understanding how AI is transforming the labor market.
“Generative AI can facilitate the exploration of other jobs without increasing the likelihood of changing companies”
In a working paper we have just published, we analyze this distinction using the 2025 Household Economics and Decisionmaking Survey (SHED) conducted by the United States Federal Reserve. The survey includes data on the use of generative AI at work and an experiment in which employees were randomly assigned to different pay scenarios: a salary freeze for one year or a reduction of 1%, 5% or 10%. Afterward, participants indicated to what extent they would actively seek another job or leave their current position in light of the assigned pay scenario. Because the different salary changes were assigned randomly, we can compare how workers react to a deterioration of their pay conditions. However, readers should note that differences between those who use generative AI and those who do not should be interpreted with caution, since the use of this technology was not randomly assigned.
The results show that hypothetical larger salary cuts increase stated intentions to search for another job or to leave the current one. Not surprising. However, we found no evidence that AI users respond differently than non-users. The main difference lies in the self-reported job-search behavior: workers who use generative AI show an approximately five percentage-point higher probability of having applied for another position in the previous year. Moreover, among users we observe a monotonic pattern: the probability of having applied for another job increases with the frequency of AI use. In contrast, we find no robust evidence that AI users have a higher probability of starting a new job or switching employers.
“We found no robust evidence that AI users are also more likely to start a new job or switch employers”
This pattern suggests that generative AI may be acting mainly on a specific stage of the labor mobility process: the job search. One possible explanation is that these tools reduce the cost of identifying vacancies, comparing alternatives, or preparing applications, facilitating the exploration of other jobs without necessarily implying better external opportunities.
The distinction between job search and actual mobility also has implications for labor market policies. The use of generative AI is spreading among the workforce much more broadly than employer demand for specialized AI skills. In fourteen OECD countries, vacancies requiring AI-related skills represented less than one percent of online job postings between 2019 and 2022, while recent data suggest that less than one percent of workers will need advanced AI-specific capabilities, such as model development or programming.
For most workers, AI seems to increase mainly the importance of digital, analytical, managerial, and problem-solving skills, rather than a need for highly specialized knowledge. If AI mainly facilitates job searching, broader use could lead workers to explore more opportunities without necessarily changing companies more often.
“If AI mainly facilitates job searching, broader use could lead workers to explore more opportunities without necessarily changing companies more often”
From this arises a possible challenge for employment and training policies. Promoting the use of AI is appropriate, but it is also necessary to ensure that the skills workers gain from using it are valued by other companies. It is important to improve information about vacancies and the competencies demanded by firms, to facilitate access to training, and to help workers better demonstrate what they can do. Thus, a greater job search will have a better chance of translating into better employment opportunities.
Hence, the challenge is not only that AI helps workers search more, but that said search ends up turning into better job opportunities.
