Chief financial officers (CFOs) across America are beginning to align on where artificial intelligence is most likely to have its earliest impact, and for many, that points to administrative work.
According to a survey of about 750 CFOs conducted by economists from Duke University and the Federal Reserve Banks of Atlanta and Richmond, executives are more likely to see AI reducing roles in routine, clerical, and office support functions than enhancing them. That expectation, however, isn’t translating into immediate job losses.
The study found that AI has had little to no measurable effect on employment so far in 2025, with most companies projecting only modest changes in the near term. Yet the reality remains different as larger firms with 500 or more employees are leaning toward trimming routine roles while keeping technical hiring steady. Smaller companies, though, view AI as a chance to grow, maintaining administrative positions and expanding their teams with skilled technical talent.
John Graham, an economist at Duke University and one of the study’s authors, told the Wall Street Journal that workers in more highly educated or specialized roles could eventually feel the impact of AI, “but probably not in 2026.”
After surveying CFOs for more than 30 years, Graham notes their perspective is particularly valuable. “CFOs are uniquely placed to understand the inner workings of their companies,” he said, “since it is their job to keep watch on how company resources are being deployed.”
What This Actually Means for Job Roles
While it is stated that AI has so far had only a modest effect on overall employment, Graham pointed out that the study focused on established companies rather than newer ones. That distinction matters because younger or smaller firms often adopt technology in ways that create new opportunities.
Higher-skill, specialized positions like engineering, advanced analytics, and other technical roles are more likely to be augmented rather than replaced, as workers in these areas can use AI to increase productivity and output, rather than compete against it.
Economists call this pattern skills-biased technological change, where new technologies tend to hollow out routine work while amplifying highly educated or trained employees. When personal computers entered offices decades ago, financial analysts, scientists, and consultants expanded their capabilities, while typists and back-office bookkeepers saw their share of jobs shrink.
A tricky part for workers is that losing a job doesn’t guarantee a spot in the new roles AI might create. Salomé Baslandze, an economist at the Atlanta Fed and one of the study’s authors, is hopeful that AI will open up new kinds of work. But she also warns that many of the positions CFOs see as vulnerable are what she calls “stepping stones” into the middle class.
Graham describes this as the divide between company types. “Small companies look at this and think, ‘This gives us opportunities to expand,’” he said. The implication is that AI’s effects won’t be the same everywhere; how a firm approaches adoption, its growth strategy, and its size all influence who feels the impact.
David Adubiina
