AI-Impacted Jobs Decline for Second Year: BLS Data Reveals 0.2% Drop Amid Broader Growth

Recent data from the US Bureau of Labor Statistics highlights a concerning trend for workers in AI-exposed occupations. From May 2024 to May 2025, employment across 18 roles most vulnerable to automation fell by 0.2%, contrasting with a 0.8% rise in the overall US labor market. This marks the second consecutive year of job losses in these sectors, raising questions about the long-term effects of artificial intelligence on employment. Below, we explore key questions about these findings, their implications, and what they mean for the future of work. Jump to Question 1, Question 2, Question 3, Question 4, Question 5, or Question 6.

What specific occupations are considered AI-exposed, and how were they selected?

The US Bureau of Labor Statistics (BLS) identified 18 occupations as AI-exposed based on their susceptibility to automation and artificial intelligence. These roles typically involve repetitive tasks, data processing, or routine decision-making that can be replicated by machine learning algorithms. Examples include telemarketers, proofreaders, data entry keyers, legal secretaries, and bookkeeping clerks. The selection criteria combined factors like task automation potential, historical employment trends, and expert assessments from labor economists. Each occupation was rated on how much of its core work could be performed by current or near-future AI systems. The list is periodically updated to reflect technological advancements. For a deeper dive into why these roles are vulnerable, see Question 3.

How does the 0.2% decline in AI-exposed jobs compare to the broader labor market performance?

The contrast is stark: while the overall US labor market expanded by 0.8% from May 2024 to May 2025, employment in the 18 AI-exposed occupations shrank by 0.2%. That difference of 1.0 percentage point is significant in relative terms. To put it in perspective, if the AI-exposed sectors had grown at the same rate as the rest of the economy, they would have added tens of thousands of jobs instead of losing them. This divergence isn't new—it mirrors the previous year's pattern, where AI-vulnerable roles also underperformed the broader market. The BLS data underscores a structural shift: the industries most likely to be disrupted by AI are bleeding jobs even as other sectors thrive. For a breakdown of which industries are hit hardest, see Question 5.

Why might AI-exposed occupations be experiencing job losses for a second year?

The second consecutive year of decline suggests systematic factors beyond normal economic cycles. Key drivers include:

• Technological maturation: AI tools like generative models and robotic process automation are becoming more reliable and affordable for businesses.
• Cost reduction pressure: Firms in sectors like legal services and accounting have strong incentives to replace human labor with software to cut expenses.
• Post-pandemic restructuring: Many companies accelerated digital transformation during COVID-19, and those changes are now permanent.
• Lack of retraining: Workers in AI-exposed roles often lack pathways to transfer to growing fields like healthcare or tech.

These forces compound each year, making it harder for displaced workers to find similar jobs. The BLS notes that the pace of job loss may accelerate if AI adoption continues to spread. For more on worker implications, see Question 6.

Are any AI-exposed occupations showing growth despite the overall decline?

Yes, not all AI-exposed jobs are shrinking. Within the 18-occupation basket, a few roles have bucked the trend. For instance, computer support specialists and user interface designers (often grouped as AI-adjacent) have seen mild gains, likely because they involve supervising or improving AI systems. Additionally, market research analysts have benefited from increased demand for data-driven insights, even as some of their routine tasks become automated. However, these pockets of growth are too small to offset the losses in larger categories like office clerks and customer service representatives. The net effect remains negative, with the decline concentrated in high-volume, low-skill roles. For a detailed list of which occupations are most affected, see Question 1.

What sectors are most affected by these AI-driven job losses?

The BLS data shows that administrative and support services account for the largest share of losses, including roles like data entry, payroll clerks, and receptionists. The legal sector is also heavily impacted, with paralegals and legal secretaries seeing significant declines due to AI-powered document review and contract analysis. Other notable sectors include:

1. Retail and wholesale trade (cashiers, stock clerks)
2. Finance and insurance (loan officers, underwriters)
3. Manufacturing (quality control inspectors)

These sectors have high proportions of routine cognitive and manual tasks. The losses are concentrated in metro areas with high office employment, such as New York and San Francisco. In contrast, industries like healthcare and hospitality (less exposed to AI) have added jobs. For strategies workers can use, see Question 6.

What do these trends mean for workers in AI-vulnerable roles?

For workers currently in AI-exposed occupations, the outlook demands proactive adaptation. The data signals that staying in the same role without upskilling is risky. Recommended steps include:

• Upskilling into areas like data analysis, AI ethics, or cybersecurity, which are less automatable.
• Pursuing education in fields requiring human judgment, such as nursing, sales management, or skilled trades.
• Mobility: Migrating to regions or sectors with lower AI exposure, such as rural healthcare or renewable energy installation.

Employers also have a role: the best companies invest in reskilling programs rather than simply laying off workers. Policymakers may need to expand safety nets and training subsidies. The BLS data is a call to action, not an inevitability—with planning, many workers can transition to thriving roles. For a full list of AI-exposed occupations, see Question 1.