AI's Toll on Jobs: 10 Key Findings from the Latest US Labor Data

The US Bureau of Labor Statistics recently released data comparing employment shifts from May 2024 to May 2025, revealing a stark contrast: while the overall labor market grew by 0.8%, employment across 18 occupations highly exposed to artificial intelligence dipped by 0.2%. This marks the second consecutive year of notable job losses in these fields, underscoring AI's accelerating impact on specific sectors. In this listicle, we break down the data into 10 critical insights, exploring which roles are most affected, why the decline matters, and what it signals for the future of work.

1. The 0.2% Decline: A Small Number with Big Implications

At first glance, a 0.2% drop over 12 months may seem insignificant, especially compared to the 0.8% rise in overall employment. However, context is crucial. The 18 AI-exposed occupations represent a narrow slice of the labor market—jobs like data entry, telemarketing, and translation that rely heavily on routine cognitive tasks. A decline in these roles, even modest, suggests that AI adoption is not just automating isolated tasks but beginning to reshape entire job categories. The persistence of this decline for a second year amplifies the warning signal: it's not a one-time adjustment but a trend.

2. Broader Market Growth Masks Sector-Specific Pain

The 0.8% overall job market expansion indicates a resilient economy, with gains in healthcare, leisure, and construction offsetting losses elsewhere. Yet for workers in AI-exposed fields, this growth offers little comfort. The divergence highlights a structural shift: AI is not causing net job losses across the economy—but it is disproportionately hitting white-collar roles that involve repetitive information processing. As noted in point 6, these occupations often lack the physical or interpersonal elements that protect other jobs from automation.

3. Second Year of Losses: A Worsening Trend

The data shows that several occupations expected to be impacted by AI experienced heavy job losses for the second year in a row. This suggests that the initial shock of automation is not evening out. Instead, it's intensifying as AI tools become more capable and cheaper. For example, generative AI and large language models have made significant inroads in areas like copywriting and customer service, roles that were earlier considered safe from full automation. The repeat nature of these losses means that workers may need to retrain or relocate permanently.

4. The 18 Occupations: Who Is Most at Risk?

While the Bureau of Labor Statistics does not list every occupation in this report, the categories often cited include: data entry keyers, telemarketers, proofreaders, translators, customer service representatives, and bookkeeping clerks. These jobs share a common feature: they involve high volumes of predictable, rule-based tasks that AI systems can perform at equal or better accuracy. For instance, translation tools have improved dramatically, reducing demand for human translators—especially for general content. This point 8 delves deeper into sector-specific impacts.

5. Geographic Distribution: Uneven Impact Across States

Though the national average shows a 0.2% decline, certain regions likely suffer more. Areas with high concentrations of administrative work—like the Washington D.C. metro area (government), or California's tech hubs (customer support)—may see steeper losses. Conversely, states with strong manufacturing and healthcare sectors may buffer the impact. Unfortunately, the Bureau's aggregated data does not break down by state, making local analysis essential for policymakers who need to target retraining programs. Without geographic granularity, the true burden on specific communities remains hidden.

6. Why AI-Exposed Jobs Are Shrinking While Others Grow

The 0.8% growth in overall employment came almost entirely from sectors requiring physical presence, emotional intelligence, or complex problem-solving—traits that current AI struggles to replicate fully. Jobs like nursing, construction, and food service depend on human touch, adaptability, and dexterity. In contrast, AI-exposed roles are often performed remotely or require minimal physical interaction, making them easier to automate. This divergence underscores a key economic principle: AI displaces tasks, not necessarily jobs—but when entire roles consist of automatable tasks, displacement becomes inevitable.

7. Comparison to Previous Year: The Trend Accelerates

Last year's data (May 2023 to May 2024) already showed a drop in these same 18 occupations. The second consecutive decline suggests acceleration rather than equilibrium. Economists initially speculated that some job losses would be temporary as firms experimented with AI. But the repeat pattern indicates that after initial trials, companies are permanently replacing human workers with AI software. For example, many call centers are now using AI voice agents for routine inquiries, reducing the need for human operators.

8. Sectors Feeling the Heat: Translation and Customer Service

Among the hardest-hit fields are translation services and customer service. Neural machine translation like DeepL and Google Translate now offers near-human quality for many language pairs, slashing demand for human translators—especially for business documents and routine correspondence. Similarly, AI chatbots handle first-line customer queries for e-commerce and tech support, with human agents only escalating complex issues. The result: fewer entry-level positions in these fields. Point 4 listed these among the 18 at-risk occupations, and the data confirms the trend.

9. The Human Cost: Wage Stagnation and Career Disruption

Beyond headcount, the job losses carry a human toll. Workers in displaced roles often face wage stagnation if they find reemployment in less specialized fields. A customer service representative who loses a job may end up in retail or gig work, earning less. Moreover, older workers may struggle to retrain, while younger ones might avoid these career paths altogether. The story is not just about numbers—it's about livelihoods. Policymakers must consider safety nets like universal retraining vouchers or expanded unemployment benefits, as outlined in many automation-impact reports.

10. Looking Ahead: What the Next 12 Months Might Bring

If current trends hold, the 18 AI-exposed occupations could see further declines of 0.3% to 0.5% in the next year. Meanwhile, new roles may emerge in AI oversight, prompt engineering, and data labeling—but these require different skills and may not absorb all displaced workers. The Bureau's data serves as an early warning: retraining and education systems must adapt quickly. Employers, too, have a role—by investing in human capital alongside technology. The 0.2% drop is a small number, but it points to a large transformation already underway.

In conclusion, the BLS data for May 2024 to May 2025 clearly shows that AI is reshaping specific job markets, even as the overall economy grows. The 0.2% decline in 18 AI-exposed occupations, now a two-year pattern, signals a structural shift that requires attention from workers, businesses, and government. While 0.2% seems minor, its persistence and the contrast with broader job growth make it a critical metric for understanding the future of work in the age of AI.