BLS Just Released AI-Adjusted Job Projections for 2024-34 — How They Compare to Our Data
The Bureau of Labor Statistics has, for the first time, explicitly factored AI into its 10-year employment projections. We compared their numbers against our AI automation risk data for 10 key occupations.
For years, government employment forecasts treated artificial intelligence as background noise — a vague "technology" factor buried in methodology footnotes. That changed when the U.S. Bureau of Labor Statistics released its 2024-34 employment projections, making this the first cycle where AI impacts are explicitly incorporated into the official U.S. job outlook.
We pulled the BLS numbers for 10 occupations that our readers check most often on aichanging.work, then stacked them against our own AI automation risk scores. The results reveal where government forecasters and AI researchers agree — and where they see the future very differently.
Where Government Data and AI Research Agree
The strongest alignment comes from occupations at both extremes.
Customer service representatives are projected to decline 5.5% over the decade according to BLS, shedding about 153,700 jobs from a workforce of 2.8 million. Our data assigns this occupation an automation risk of 80/100 by 2025, classifying it firmly in the "automate" category. Both sources agree: AI chatbots and automated support systems are replacing human agents at scale. BLS specifically calls out AI as a factor dampening demand here.
Bookkeeping clerks tell a similar story. BLS projects a 5.8% decline (about 94,300 fewer jobs), and our risk score sits at 66/100 with the highest-impact task — recording transactions — already at 80% automation. Both datasets point to the same conclusion: routine financial record-keeping is being absorbed by AI-powered accounting software.
Registered nurses represent the other end of the spectrum. BLS projects 5% growth, and our automation risk stands at just 12/100 for 2025. Physical patient care, clinical judgment, and emotional support remain firmly human tasks. AI assists with documentation and diagnostics, but it does not replace bedside nursing.
Accountants and auditors show moderate agreement. BLS projects 4.6% growth (about 72,800 new jobs), while our data shows a moderate automation risk of 50/100. This is the "augmentation" sweet spot — AI handles tax prep and data entry while creating demand for professionals who can interpret AI-generated insights, advise clients, and navigate complex regulations.
The Surprising Gaps
Two occupations stand out with significant discrepancies between BLS projections and our AI risk assessments.
Computer programmers show the most interesting divergence over time. Our data currently lists the BLS projection at -11%, but the latest 2024-34 BLS figure is actually -6.0% (down from -9.6% in the 2023-33 cycle). This is notable: even as AI code generation tools have become dramatically more capable, BLS has actually moderated its decline forecast. One explanation is that the productivity boost from AI tools like GitHub Copilot is creating new programming demand even as it automates existing tasks. Our automation risk score of 70/100 suggests heavy transformation, but the BLS numbers hint that "transformation" does not always mean "elimination."
Administrative assistants present the starkest mismatch. Our data carries a -10% BLS projection, but the actual 2024-34 BLS figure for the aggregate secretaries and administrative assistants category shows "little or no change." However, BLS specifically identifies nonmedical secretaries (SOC 43-6014, exactly our occupation code) as an AI-impacted subcategory where employment is "projected to decline or show little change." The aggregate number is propped up by medical secretaries, where healthcare demand offsets AI productivity gains. Our -10% figure likely reflects the nonmedical subcategory more accurately, but we should update it to align with the latest BLS detailed data.
What BLS Tells Us About AI and Jobs That Researchers Miss
The BLS methodology brings something that academic AI research often lacks: industry-level demand context. A job can have high AI exposure and still grow if the underlying industry is booming. Software developers are the prime example — our data shows exposure reaching 68% by 2025, yet BLS projects 15% growth because demand for AI development itself is driving massive hiring.
Conversely, jobs can have moderate AI exposure but still decline if their industry is shrinking for other reasons. The BLS framework reminds us that AI is one force among many — demographics, regulation, trade policy, and consumer behavior all shape employment outcomes.
The BLS also introduces an important nuance about replacement versus augmentation. For lawyers (BLS: +4% growth, our risk: 30/100), AI is making legal research and document review faster but is not replacing the judgment, client relationship, and courtroom advocacy that drive demand for lawyers. For paralegals (BLS: ~0% change, our risk: 50/100), the picture is bleaker — the tasks AI automates (research, document review) are a larger share of the paralegal role.
What This Means for Your Career
If you are checking our site to understand how AI might affect your job, the BLS data adds an important layer of confidence.
If both sources agree your occupation is growing (software developers, nurses, accountants), you can plan with reasonable confidence that demand for your skills will remain strong — though the nature of the work will change significantly.
If both sources agree your occupation is declining (customer service, bookkeeping), the signal is clear. Upskilling into adjacent roles that combine domain knowledge with AI fluency will be critical. A customer service veteran who can train and manage AI chatbot systems has a far brighter outlook than one waiting for the decline to reverse.
If the sources disagree, pay attention to why. Computer programmers face high AI exposure but moderating decline forecasts — suggesting that learning to work with AI tools may be more important than fearing replacement by them.
For detailed automation risk scores, task-level breakdowns, and career recommendations for your specific occupation, visit your occupation page on our site. We are updating our BLS projection data to reflect these latest 2024-34 figures.
This analysis compares BLS Employment Projections (2024-34) published August 2025 with aichanging.work automation risk data derived from Anthropic, Eloundou et al. (2023), and Brynjolfsson et al. (2025). BLS data accessed March 2026. AI-assisted analysis.