What jobs will AI replace? The official list for 2026

AI will replace the jobs that are almost entirely routine information work first: data entry, typing, filing, payroll processing, scripted phone calls and simple customer questions. The U.S. Bureau of Labor Statistics (BLS) projects office and administrative support to lose 752,100 jobs by 2035, more than any other group, and names AI-powered automation as one of the reasons.

But the official data also shows something the scary headlines skip: most jobs that are highly exposed to AI are not disappearing. Software developers, data scientists and accountants are all rated among the most exposed, and all are projected to grow. What AI replaces is tasks. Whole jobs go only when those tasks are nearly all there is. Below is the list of jobs AI is most likely to replace, the numbers behind it, and how to tell where your own job stands.

The list: jobs AI is most likely to replace

In August 2026 the BLS published its first official AI exposure ratings for 831 occupations, alongside its job projections for 2025 to 2035. It combined five research measures, including real-world AI usage data from Anthropic and Microsoft, and sorted every occupation into one of four roughly equal groups: low, moderate, high or very high exposure.

We filtered that data for the jobs that are both rated very high for AI exposure and projected to shrink fastest. These are the jobs where AI overlap and falling demand line up:

The pattern is clear. Almost every job on the list turns information from one form into another: paper into a database, a script into a phone call, timesheets into a payroll run. That is exactly what today’s AI and automation software does well.

Two other jobs are worth knowing even though the BLS rates them one step lower, at high exposure: bank tellers are projected to fall 13.2% and cashiers 6.5%, a loss of about 200,600 cashier jobs, the largest numeric drop in this part of the data. Self-checkout, apps and online banking are doing much of that work.

The biggest job losses in raw numbers

Percentages can hide scale. A small occupation can shrink fast without many people losing work, while a large one shrinking slowly affects far more people. Ranked by the number of jobs, the very highly exposed occupations losing the most by 2035 are:

• General office clerks: −156,200 jobs (−6.0%)
• Customer service representatives: −141,800 jobs (−5.3%)
• Secretaries and administrative assistants (outside legal, medical and executive roles): −114,100 jobs (−6.0%)
• Bookkeeping, accounting and auditing clerks: −85,600 jobs (−5.6%)
• Data entry keyers: −33,600 jobs (−25.5%)
• Computer user support specialists: −26,200 jobs (−3.5%)
• Payroll and timekeeping clerks: −25,400 jobs (−15.9%)
• Insurance claims adjusters, examiners and investigators: −20,600 jobs (−5.5%)

Customer service shows how this plays out. The BLS expects less demand for customer service representatives as self-service systems, social media and apps let customers handle simple tasks themselves, and says automated systems will gradually take on more. Yet it still expects about 289,500 openings a year in the field, all of them to replace workers who retire or move on. Companies keep people for complex problems, such as refunds and insurance questions. A shrinking job is not the same as an empty one.

What jobs will AI replace first?

The first jobs to go are the ones where a single, routine task fills most of the week and the result is easy to check. Real-world usage data points to the same group.

Anthropic measured which work tasks its Claude models are actually used for, giving more weight to fully automated use than to help with a task. In its March 2026 report, computer programmers topped the list with 75% of their tasks covered, followed by customer service representatives. Data entry keyers had 67% of their tasks covered, since reading documents and entering data is now heavily automated.

Microsoft analyzed 200,000 anonymized conversations with its Copilot assistant and scored 785 occupations on how much of their work AI was used for, and how successfully. The top of its list is dominated by language work: interpreters and translators ranked first, writers and authors third and customer service representatives seventh, with telemarketers, reporters, technical writers and proofreaders close behind.

Here is the important twist. Interpreters and writers rank near the very top for AI use, yet the BLS does not project them to collapse: interpreter jobs are expected to grow 2.0% and writer jobs to hold roughly flat (−0.3%). Heavy AI use speeds up the work; it doesn’t automatically remove the person doing it. The jobs that actually shrink are the ones where AI usage replaces the work rather than assisting it.

That is also what Stanford’s Digital Economy Lab found in payroll data covering millions of U.S. workers. It saw no sign of widespread job displacement across the economy. But employment of workers aged 22 to 25 in AI-exposed jobs is now 19% below where it would be had it kept pace with less-exposed peers, mainly because companies are hiring fewer young people. The declines are concentrated in jobs where AI substitutes for human tasks.

How we score these jobs

The BLS data is about whole occupations. Our free test scores the mix of tasks in a real person’s week instead, from data and writing to hands-on and people work. Run on typical task mixes, our model puts the same jobs at the top:

A typical data entry clerk scores 96 out of 100 and a customer support agent 88, both in our “very high” band. Bookkeepers and paralegals score 84. A typical electrician scores 19 and a nurse 16, because most of their week is physical, in-person work that AI can’t reach. You can see how the score works on our method page.

What jobs will AI replace by 2030?

For the next five years, the best evidence is what employers themselves expect. The World Economic Forum’s Future of Jobs Report 2025 is based on a survey of large employers worldwide. They expect the biggest declines in numbers among clerical and secretarial workers, including cashiers, ticket clerks, administrative assistants and executive secretaries. The fastest-declining roles include postal service clerks, bank tellers and data entry clerks.

The same employers expect technology roles such as big data specialists, AI and machine learning specialists and software developers to grow fastest in percentage terms. Frontline jobs, such as farmworkers, delivery drivers and construction workers, are expected to add the most jobs in total, and care jobs such as nursing are also expected to grow strongly.

How many jobs will AI replace by 2030?

There is no trustworthy count of jobs lost to AI alone, and you should be wary of any article that gives one. The World Economic Forum’s estimate covers all the big forces reshaping work, from AI and automation to the cost of living and the energy transition. It expects 92 million jobs worldwide, about 8% of today’s total, to be displaced between 2025 and 2030, and 170 million new ones to be created, a net gain of 78 million.

The U.S. official outlook is similar in direction. The BLS projects total employment to grow 3.5% from 2025 to 2035, adding 5.9 million jobs, even after factoring in AI. Its own release says AI is expected to support demand in some occupations while productivity gains may dampen demand in others.

What jobs will AI replace in the next 10 years?

The BLS ten-year outlook runs to 2035, and it groups the pressure into three areas:

• Office and administrative support is projected to decline fastest of all 22 major groups (−4.0%), as automation tools, including those powered by AI, take over more of the workflow.
• Sales is projected to lose jobs (−1.4%), as e-commerce grows and AI tools are built into selling.
• Arts, design, entertainment and media jobs may see less demand because generative AI automates repetitive tasks. Graphic designers (−1.7%), editors (−1.1%) and desktop publishers (−14.5%) are projected to shrink.

Within those groups, the jobs at greatest risk are the ones on the list above. Beyond 2035 nobody can forecast specific jobs credibly, including for searches like “what jobs will AI replace by 2040 or 2050.” What we can say is which kinds of work are hardest to automate, and the same answer comes up in every study.

Jobs AI is less likely to replace

Highly exposed jobs that are still growing tend to have one thing in common: the person decides what to build, checks the result and answers for it. The BLS projects strong growth for several very highly exposed occupations:

• Data scientists: +34.6%, one of the 10 fastest-growing jobs in the country
• Computer and information systems managers: +15.8%
• Software developers: +10.2%, adding 174,700 jobs, even as computer programmers shrink
• Management analysts: +10.1%
• Accountants and auditors: +5.0%, as we explain in Will AI replace accountants?

The split between programmers and developers is the clearest example in the data. Writing code to a specification is shrinking; designing systems and deciding what software should do is growing.

At the other end, the least exposed work is physical and in person. In Anthropic’s data, about 30% of workers are in jobs where AI coverage was too rare to measure. Microsoft gives a score of zero to jobs such as orderlies, floor sanders and dredge operators, and nursing assistants rank 745th of 785. The BLS expects healthcare to account for almost a third of all new U.S. jobs by 2035. For the full ranking of safe careers, see the 25 most AI-proof jobs. These three careers from our index show the kind of work that combines low exposure with good prospects:

How to stay ahead if your job is on the list

If your job is on the list, you don’t need to panic, but you should start moving. Openings will still come up for years as people retire, and your experience is worth more than you think. The goal is to shift your week toward the tasks AI can’t do. Here is how each type of task holds up:

1. Find out which of your tasks are exposed. Your job title is a rough guide; your week is the real one. Add up the hours you spend on data handling, routine writing and scripted calls.
2. Use AI on the routine part, and own the check. People who can direct AI tools and catch their mistakes are taking work from people who can do neither.
3. Move toward exceptions, judgment and people. Every shrinking job on this list keeps a core of complex cases: the angry customer, the messy account, the unusual claim. Become the person who handles those.
4. Pick a neighboring role that is growing. Office clerks move into project coordination and operations; bookkeepers into advisory accounting; customer service agents into account management, sales or support for AI systems.
5. Consider skills AI can’t reach. Skilled trades and healthcare offer some of the safest long-term demand, often with paid training.

Our guide to the skills AI can’t replace turns this into a 90-day plan. And if you work in law, see Will AI replace lawyers? for how the same pattern plays out between lawyers, paralegals and legal secretaries.