15 Jobs Most Affected by Generative AI — U.S. Labor Statistics Data and How to Future-Proof Your Career

"Could AI take my job?" If that vague worry has crossed your mind, you are far from alone.

The World Economic Forum's Future of Jobs Report 2025, published in January 2025 and based on a survey of more than 1,000 employers worldwide, expects that between 2025 and 2030 about 170 million jobs will be created and about 92 million displaced, for a net gain of about 78 million. In total, work is not disappearing.

But the jobs that shrink and the jobs that grow are not the same jobs. This article focuses on one question — which occupations are most affected — and answers it occupation by occupation with two primary sources: the ten-year projections the U.S. Bureau of Labor Statistics (BLS) published in August 2026, and data on how AI is actually being used. What makes a job likely to last is covered in Jobs That Survive the AI Era, the gap between junior and senior workers in Veterans vs. Juniors, and the gap between white-collar forecasts and the actual data in Will AI Eliminate White-Collar Jobs?

Why AI Job Loss Feels Real Now

People have talked about automation taking jobs for decades, but generative AI changed the picture. There are three main reasons.

1. Generative AI now handles knowledge work

Large language models (LLMs) such as ChatGPT and Claude can write, translate, program and analyze data — work that used to be considered uniquely human. Earlier waves of automation mostly hit factories and warehouses; this one reaches the desk work of office and professional jobs.

2. Adoption has become cheap

Deploying AI once required heavy investment. Today, a company can build capable AI into its workflows for the price of API usage. Now that AI is realistic for small and mid-sized businesses too, the range of affected work is widening.

3. Employers are writing AI into their headcount plans

In the same WEF survey, 40% of employers said they expect to reduce their workforce where AI can automate tasks. At the same time, two-thirds plan to hire people with specific AI skills. In practice, the jobs a company cuts and the jobs it adds are moving at the same time.

15 Jobs Most Affected by AI — According to U.S. Labor Statistics

Many "X% chance AI replaces this job" rankings online never say where the number came from or how it was calculated. Here we use figures with a clear publisher. On August 27, 2026, the U.S. Bureau of Labor Statistics (BLS) published its 2025–2035 employment projections, along with the factors behind each occupation's projection. Below are the occupations for which BLS names AI as a factor, sorted by projected change.

Three things stand out in this table.

• Even "declining" jobs mostly shrink by less than 10%. The steep drop is data entry (−25.5%). Customer service is −5.3%, and paralegals and writers are roughly flat. But because the all-occupation average is +3.5%, flat still means shrinking relative to the economy as a whole.
• The reason is not only replacement. Many of BLS's explanations are about efficiency — as with claims adjusters or technical writers, AI lets each person handle more work, so the same output needs fewer people. The occupation does not vanish; the headcount it needs goes down.
• Within IT, the direction splits. Computer programmers are projected at −7.3%, but software developers at +10.2%. Jobs made up only of writing code shrink; jobs that also cover design and requirements grow.

For some occupations often said to be "doomed by AI," BLS paints a different picture. Interpreters and translators: +2.0%. Accountants and auditors: +5.0%. In neither case does BLS list AI as a factor. And the large declines for bank tellers (−13.2%), cashiers (−6.5%) and bookkeeping clerks (−5.6%) are, in the BLS table, not attributed to AI either — online banking, self-checkout and other automation that predates generative AI are doing the work there.

Alongside the projections, BLS also released AI exposure categories that rank occupations into four levels of relative exposure to AI. It is worth remembering BLS's own caveats: high exposure "does not necessarily mean employment will decline," and low exposure does not mean an occupation is safe. These are U.S. figures and do not transfer directly to other countries.

What Makes a Job Hard for AI to Replace

So what do jobs that resist AI have in common? Four traits stand out.

What the four share is that the core of the job is something other than processing set procedures through text or data. The occupations on the declining side of the table above all spent much of their time on entry, transcription, routine customer handling or routine documents. When you assess your own job, asking "what share of my day is procedural document or data processing?" gets you closer to reality than looking at the job title.

Why particular occupations tend to last — with BLS figures for the growing ones (nursing, home care, electrical work, information security and more) and what it takes to get into each — is covered in detail in Jobs That Survive the AI Era.

What Usage Data Shows — Where AI Is Already Doing the Work

The BLS figures are a projection for the next ten years. So which occupations' tasks is AI actually being used for today? One study measures exactly that from usage data.

Anthropic's "observed exposure"

In March 2026, AI company Anthropic mapped work-related use of its own model, Claude (usage data from August and November 2025), onto occupational tasks and measured how much of what AI could theoretically do is actually being done with it, weighting automation-style use more heavily. Because this is an AI vendor's own usage data, it should be read with that in mind — but it is one of the few primary sources that reports actual use by occupation.

The interesting part is how this lines up with BLS. The same study reports a slight relationship: for every 10-point increase in observed exposure, the BLS growth projection is 0.6 points lower. Programmers, customer service and data entry — the top of Anthropic's list — were also on the declining side of the BLS table. Two independent data sources point at the same occupations.

A company example: Klarna's customer service

One visible example of change at the level of a whole function is Swedish fintech Klarna. In a February 2024 announcement, the company said its AI assistant handled 2.3 million conversations — two-thirds of its customer service chats — in its first month, doing the equivalent work of 700 full-time agents. That is the company's own figure, not an independent measurement. Its direction matches customer service ranking high in both the BLS projections and Anthropic's data.

5 Strategies to Thrive in the Age of AI

If this has made you uneasy, remember that the goal is to prepare, not to panic. Here are five concrete strategies.

Strategy 1: Be the one who uses AI well

This matters most. As the table showed, many of BLS's reasons come down to efficiency: AI raises how much one person can handle. If a team needs fewer people, the ones who stay are those who use AI to increase their own output.

Use tools like ChatGPT and Claude every day, and keep asking how they fit your own work. In the WEF survey, two-thirds of employers plan to hire people with AI skills. To get more out of Claude, see our guide to Claude's three tabs.

Strategy 2: Build your human skills

Empathy, negotiation, leadership, team building — the more routine work AI takes on, the larger the share of your job these interpersonal skills become.

AI can take over technical tasks, but the ability to move people and build trust is hard to substitute. Investing in interpersonal skills pays off for a long time.

Strategy 3: Go deep in a specialty

AI is good at quickly using knowledge that has been widely written down. That means people with deep expertise and field experience in a specific area are harder to replace.

Aim to have something you are the best at, rather than being moderately good at everything. The more a field depends on experience and judgment that are not written down, the more human experience counts.

Strategy 4: Combine skills

One skill alone may feel fragile, but combining skills from different fields creates value AI finds hard to copy.

Think "marketing × data science," "design × programming," or "healthcare × AI." Combine skills to build a position that is yours alone. How the gap between office workers who use AI and those who don't is widening is covered in How AI Widens the Skill Gap.

Strategy 5: Keep learning

AI changes quickly; what was common knowledge last year may not hold this year. What matters is less any specific skill than the ability to keep learning. The WEF survey expects 39% of workers' existing skills to be transformed or outdated by 2030.

Online courses, books and communities offer plenty of ways to learn. Instead of fearing change, build the ability to adapt to it.

Conclusion: Prepare, Don't Panic

AI will change work — that is unavoidable. But looking back, the Industrial Revolution and the internet both eliminated old jobs while creating new ones.

Three things matter most:

• Assess your risk by task, not job title: the occupations where BLS cites AI are heavy on data entry, transcription, routine customer handling and routine documents
• Treat AI as a tool, not an enemy: much of the decline comes from efficiency reducing headcount. Those who stay are the ones who use AI to do more
• Start today: the change plays out over a decade, and the earlier you prepare, the better

Instead of fearing that AI will take your job, look for a way of working alongside it — and take the first step today.

FAQ