“While AI may have started to change our workflows, its role in explaining the recent slowdown in job growth is overstated,” Vanguard’s latest analysis concludes a finding that directly challenges the prevailing fear that artificial intelligence is rapidly erasing human employment. In fact, the data shows that occupations most exposed to AI automation are currently expanding faster than the rest of the labor market, with wages rising at an even sharper pace.
Vanguard examined roughly 140 occupations deemed highly vulnerable to AI automation roles such as office clerks, typists, HR assistants, law clerks and data scientists because they involve a high share of tasks that AI systems could theoretically perform autonomously. These positions, often cited as the likeliest to shrink as AI adoption accelerates, have instead grown by 1.7% between mid-2023 and mid-2025 , compared with a 1% increase in the pre-Covid period from 2015 to 2019. By contrast, job growth in all other occupations slowed over the same post-Covid period.
The wage picture is even more striking. Real wages in AI-exposed roles, adjusted for inflation, rose 3.8% post-Covid , compared with just 0.1% pre-Covid. Jobs less exposed to AI saw smaller gains, from 0.5% to 0.7%. If AI were truly damaging the job market at scale, economists note, the impact would be visible in shrinking paychecks yet the opposite is occurring.
This pattern echoes historical precedents in technology adoption. Economists such as Anders Humlum point out that transformative technologies from steam power to electricity to computers often take decades to produce large-scale economic effects. In early phases, productivity gains can spur new demand, creating more work rather than less. Box CEO Aaron Levie recently illustrated this with software engineering: “If you lower the cost of building software, you have 10 times the use cases for software. That’s the part that everybody misses.”
Early-stage AI deployment is still uneven and often experimental. According to the McKinsey State of AI 2025 report, nearly two-thirds of organizations have yet to scale AI across the enterprise. Many deployments focus on augmentation rather than full automation, enabling workers to shift from rotating tasks to higher-value activities. Harvard’s Digital Data Design Institute found that AI-enabled teams can outperform individual AI-equipped workers in innovation quality, suggesting that integrating AI into collaborative workflows can enhance rather than replace human contributions.
For entry-level workers, the picture is more complex. While some Federal Reserve anecdotes note modest AI-driven reductions in hiring, Vanguard’s internal data shows that the share of workers aged 21 to 25 enrolling in its 401(k) plans remains high. Harvard’s research also reveals that lower-skilled workers see the largest performance boosts up to 43% when equipped with AI, though this can create a double-edged sword: if AI handles junior tasks better, senior staff may delegate less, limiting on-the-job training for newcomers.
Economic conditions have also played a significant role in shaping the current labor market. Analysts like Jing Hu argue that the recent slowdown in junior hiring began before widespread AI adoption, driven by aggressive Federal Reserve rate hikes starting in 2022, pandemic aftershocks, and cost-cutting measures unrelated to automation. This mirrors past downturns, where entry-level hiring was disproportionately cut and often failed to recover to pre-shock levels.
For professionals and economists tracking AI’s impact, the takeaway is that the technology’s current limitations from model inaccuracies to incomplete integration into workflows are preventing widespread displacement. However, as AI systems improve, certain roles such as customer service representatives, paralegals, and even economists may face declining demand. The challenge ahead lies in strategic adaptation: upskilling to meet rising baseline skill requirements, designing processes that leverage AI’s strengths while preserving human creativity, and ensuring that productivity gains translate into broadly shared economic benefits.
The early evidence suggests AI adoption can add work before it removes it, and in doing so, can raise pay for those who integrate it effectively. Whether this trend endures will depend on how quickly AI capabilities advance, how organizations redesign roles around them, and how workers position themselves to thrive in an evolving employment landscape.
