AI's impact on jobs remains modest so far, but young workers face headwinds

A growing body of research is beginning to test the "AI jobs apocalypse" narrative against real labor market data. Dario Amodei, CEO of Anthropic, has publicly predicted that AI could eliminate half of all white-collar jobs and push unemployment to 20 percent—a claim that has amplified fears across media and policy circles. Yet when economists examined unemployment trends from 2022 onward, separating workers into quintiles by their occupational exposure to AI disruption, the pattern contradicted this narrative. The unemployment rate for the top quintile of AI-exposed workers rose by 0.77 percentage points, while the rate for the least-exposed workers rose by 0.85 percentage points—a broader softening of the labor market rather than AI-specific displacement.

Employment in highly AI-exposed occupations has remained remarkably stable. Software developer positions, among the most exposed roles, have seen job postings grow faster over the last year than for other occupations. Firms adopting enterprise AI systems saw employment grow by 10 percent in the two years following adoption. When companies have announced layoffs citing AI as a factor, labor economists and industry figures have expressed skepticism about the claims, noting that some layoffs appear driven by a desire to free up cash for AI investments or to reduce headcount after pandemic-era over-hiring. HR executives report that AI's impact is more evident in role consolidation and hiring avoidance in automatable roles than in mass terminations.

Yet a sharper picture emerges for junior workers. Recent graduates are facing their most challenging job market in years, with unemployment reaching 5.6 percent in early 2026, up 1.6 percentage points from three years prior. Research by Brynjolfsson, Chandar, and Chen documented a notable decline in employment among early-career workers in software development and customer service since ChatGPT's November 2022 launch, while employment among older workers in the same occupations remained stable or continued to grow. The authors describe these young workers as "canaries in the coal mine," the first to experience labor market disruption. Similar negative effects on early-career hiring have been identified in the U.S. and the U.K. beginning in 2022.

However, causation is harder to establish than headlines suggest. The Federal Reserve began aggressively raising interest rates in March 2022, several months before ChatGPT's public release in November, and two papers find that hiring in AI-exposed occupations began declining after this monetary policy shift but before AI tools became widely available. The pandemic-driven shift to remote work, which can slow on-the-job learning, has also eroded demand for junior hires. Research shows hiring in remote-friendly occupations began skewing toward more experienced workers after the pandemic. When Brynjolfsson and coauthors added new statistical controls, employment declines among entry-level workers became less pronounced until 2024, by which point both AI adoption and model capabilities had advanced significantly. The article emphasizes that isolating AI's specific impact from competing macroeconomic shocks remains an active and challenging area of research.

On the productivity side, experimental evidence is mixed but leans positive. In controlled settings, generative AI tools like chatbots and coding assistants have disproportionately benefited less experienced and lower-performing workers. A study of an AI assistant deployed to customer service agents in a large call center found it increased overall productivity by 15 percent, with novice and less-skilled agents seeing a 30 percent improvement in issues resolved per hour, while highly skilled agents saw no improvement and their response quality fell slightly. GitHub Copilot allowed software developers to complete tasks 56 percent faster, with gains concentrated among less-experienced programmers, though other studies found more modest gains ranging from 10 percent to 30 percent depending on deployment context. ChatGPT access reduced writing time for workers of all abilities and improved writing quality among lower-ability writers. Young lawyers using AI tools completed legal work faster, and AI medical scribes showed some evidence of accelerating note-taking, though effects were small and required physician oversight. However, AI's real-world impact is complicated by what researchers call its "jagged" capabilities—performance can be strongly positive or negative depending on the specific task. When an AI assistant was deployed to help Kenyan entrepreneurs, less-skilled entrepreneurs posted lower revenues and profits than those who grew their businesses without it, because they were more likely to act on generic AI advice detrimental to their specific situations, while better performers extracted more tailored suggestions.

The article presents a fundamental disconnect between the apocalyptic narratives dominating AI discourse and what empirical data currently shows. Dario Amodei's prediction of 50 percent white-collar job loss and 20 percent unemployment has become a touchstone for AI fears, yet aggregate unemployment data does not support immediate mass disruption. The unemployment rise for the top quintile of AI-exposed workers (0.77 percentage points since 2022) is actually slightly lower than for the least-exposed quintile (0.85 percentage points), a reversal of what an "AI jobs apocalypse" scenario would predict.

However, the data reveal a more nuanced and troubling story for a specific cohort: recent graduates and junior professionals. Researchers including Brynjolfsson, Chandar, and Chen found that employment among early-career workers in software development and customer service declined notably after ChatGPT's November 2022 launch, while older workers in the same occupations remained stable or grew. The unemployment rate for new graduates reached 5.6 percent in early 2026, up 1.6 percentage points from three years prior. This pattern suggests AI may be creating concentrated disruption in entry-level hiring—roles that historically involved routine research, analysis, and writing tasks that generative AI can now perform—even as broader labor market effects remain muted.

Complexity arises in attributing this decline purely to AI. The Federal Reserve began aggressively raising interest rates in March 2022, months before ChatGPT's public release, and several papers show hiring in AI-exposed occupations began declining after this monetary policy shift. Remote work, which accelerated during the pandemic, also appears to have eroded the demand for junior hires by slowing on-the-job learning. When researchers added new controls for these confounding factors, employment declines among entry-level workers became less pronounced until 2024—by which time both AI adoption and model capabilities had advanced substantially, making direct AI causation more plausible. The article underscores that isolating AI's impact from macroeconomic shocks remains empirically challenging and an active area of research.

No comments yet. Be the first to share your thoughts!