Customer service, secretaries, and sales positions have shrunk for two consecutive years: the latest statistics from the U.S. Bureau of Labor Statistics show that AI is replacing these 17 occupations

U.S. Bureau of Labor Statistics (BLS) data shows that 18 “highly AI-exposed” occupations account for approximately 10 million jobs. From 2024 to 2025, excluding medical exceptions, 17 of these occupations have declined by 1.6% for the second consecutive year.
(Background: Cerebras’ market cap surged to $95 billion behind “Trump’s eldest son’s fund” participating in two rounds of funding)
(Additional context: Anthropic report: By 2028, in the AI dominance race, the U.S. risks losing its computational advantage to China if it doesn’t maintain leadership)

Within a year, over 130k customer service representatives have disappeared from the U.S. job market, with AI quietly filling the roles they once held. The U.S. Bureau of Labor Statistics (BLS) has marked 18 occupational categories as “highly AI-exposed”: meaning the tasks performed in these jobs are currently reasonably replaceable by AI systems, either partially or fully.

According to BLS annual wage statistics, these 18 “highly AI-exposed” occupations total about 10 million jobs. From May 2024 to May 2025, overall employment decreased by 0.2%, while U.S. total employment grew by 0.8% during the same period, indicating a divergence.

Bloomberg further analyzed: excluding the special case of “medical secretaries and assistants” (which saw increased hiring due to the high growth of the healthcare industry), employment in the remaining 17 occupations declined by 1.6% between 2024 and 2025, marking the second consecutive year of similar contraction.

Looking at the specific numbers from 2024 to 2025, the three most impacted occupations are:

• Customer Service Representatives: decreased by 130,180 jobs, a 4.8% decline
• Secretaries and Administrative Assistants (excluding healthcare, legal, and executive roles): decreased by 31,030 jobs, a 1.8% decline
• Wholesale and Manufacturing Sales Representatives (excluding technical/scientific roles): decreased by 28,670 jobs, a 2.3% decline

Over a longer timeline, from May 2022 (the last full data point before ChatGPT’s emergence) to now, the occupations with the most significant cumulative declines are: credit authorization officers, auditors, and clerks, each down 26.2%; radio announcers and broadcast DJs, down 20.8%; and sales engineers, down 13.2%.

The Divide Between “AI Substitution” and “AI Assistance”

To understand why this contraction mainly affects jobs like customer service, secretaries, and auditors, we need to clarify two concepts.

“AI substitution” jobs refer to roles where AI can directly perform core tasks: for example, answering customer calls, processing refunds, replying to standard questions, reviewing credit files, or inputting data into forms. These jobs share common features: predictable inputs, standardized outputs, and process documentation. AI doesn’t need to understand context—just enough training data.

“AI assistance” jobs are different; AI acts as a tool to help humans do their work faster and better: for example, doctors using AI to read pathology slides, lawyers quickly organizing case precedents with AI, engineers generating test code with AI. The core judgment still relies on humans; AI just accelerates the process.

Goldman Sachs’ earlier April 2026 report also provided a more detailed snapshot: AI is expected to eliminate about 16,000 U.S. jobs per month, replacing roughly 25,000 positions every four weeks, but simultaneously augment about 9,000 roles.

This isn’t a sudden collapse but a slow, structural replacement happening month by month.

Who Is Most Deeply Impacted? Young Generations Facing Narrowest Entry Points

Bloomberg highlights a structural paradox: this wave of AI substitution is concentrated in roles traditionally seen as “entry-level” career starters.

Customer service, data entry, billing, basic financial analysis, entry-level software development—these are the first jobs for many young people gaining initial work experience. Goldman Sachs also notes that Generation Z is the most affected, as they are entering fields where AI substitution is most efficient.

This creates a scenario unprecedented in decades of technological change: past waves of automation typically started by replacing mid-to-high-level repetitive technical jobs, leaving entry-level roles for newcomers to step into. But this time, entry-level positions are shrinking first.

Industries most affected include call centers, content moderation, legal support administration, billing, and early-stage software development support roles. These fields share reliance on document processing, rule-based judgment, and standardized communication—precisely the areas where large language models (LLMs), AI systems that process and generate text, excel.

Currently, opinions on AI replacing human jobs fall into two camps: one believes this trend will only accelerate; the other argues that increased productivity will create more new jobs. What do you think?