How Is AI Changing Jobs? Task by Task, and Entry Level First

14 min read

How Is AI Changing Jobs Right Now

The big picture is calmer than the headlines. The Yale Budget Lab compared the 33 months after ChatGPT came out with the early years of the personal computer and the internet. Its researchers found the US mix of jobs changing only a little faster than in those earlier waves, and the speed-up began before AI arrived. Across the whole economy, they saw stability rather than major disruption.

Now zoom in. Researchers at Stanford's Digital Economy Lab tracked payroll records for millions of US workers through June 2026. They found no sign of economy-wide job losses, but one group stood apart. Employment of workers aged 22 to 25 in the jobs most exposed to AI sits 19% below where it would be had it kept pace with less-exposed peers.

So both stories hold at once. AI changing jobs looks slow from a distance and fast at the entry level. If you have ten years in a field, the change mostly shows up as different work on your desk. If you are trying to land your first job in that field, it can show up as a posting that never opens.

The real shift: tasks change before whole jobs do

Every job is a set of tasks. A paralegal researches, drafts, files, tracks deadlines, and talks to clients. AI is good at some of those and poor at others. A few tasks shrink, the person does more of what remains, and over a few years the role means something different, even if the title stays put.

The International Labour Organization agrees. Because most jobs are made of tasks that still need a person, its 2025 study of generative AI exposure concludes that reshaping jobs, rather than wiping them out, is the most likely effect.

Is this change good or bad for workers

It depends on which tasks leave. When AI takes the dull parts and you keep the judgment, many people find the job gets better. When it takes the parts that used to train beginners, the ladder loses its bottom rung. The Stanford team saw employment fall in jobs where AI mostly does the work in place of the person, and hold flat or rise in jobs where it mostly helps the person do the work. The technology is the same. What differs is how the work gets split.

Which Jobs AI Is Changing the Most

Role by role, the clearest pattern is that office work built on words, numbers, and screens is the most exposed.

The roles with the highest share of automatable tasks

Pew Research Center sorted US jobs by how much of their core work AI could do. It found 19% of workers in the most exposed jobs and 23% in the least exposed. Budget analysts and data entry keyers ranked high. Barbers and childcare workers ranked low. The ILO's global index points the same way: clerical work carries the highest exposure of any group.

The World Economic Forum's Future of Jobs Report 2025, built on a survey of more than 1,000 large employers, lists the roles they expect to shrink fastest by 2030. Near the top are cashiers and ticket clerks, administrative assistants and executive secretaries, data entry clerks, bank tellers, and accounting, bookkeeping, and payroll clerks.

What these jobs share is a week spent moving information from one form into another: a receipt into a ledger, a request into a form, a rough draft into a clean one. The more of your week goes to reshaping information, the more of it AI can take on.

How Is AI Changing Jobs in Each Sector

Here is that split across seven sectors.

Where the disruption already shows up in hiring data

Forecasts are guesses. Hiring data is what already happened, and so far it says the change shows up first in who gets hired. In the Stanford payroll data, workers aged 22 to 25 in the most exposed jobs sit 19% below trend, while experienced workers in the same jobs show no gap. That drop comes from companies hiring fewer beginners for the tasks AI now handles, not from layoffs. Here is how the main studies line up side by side.

Employers have said the same thing out loud:

• IBM, May 2023. Chief executive Arvind Krishna paused hiring for back-office roles and expected AI to replace about 7,800 of them over five years.
• Klarna, February 2024. The payments company said its AI assistant handled two-thirds of customer service chats, the work of 700 full-time agents.
• Shopify, April 2025. Chief executive Tobi Lütke told teams to show why AI cannot do the work before asking for more headcount.
• Salesforce, September 2025. Chief executive Marc Benioff said support staff fell from about 9,000 to 5,000 as AI agents took half of customer conversations.

Most of these moves work through hiring: a pause, a headcount rule, a role left empty.

How Is AI Changing Jobs for Mid-Career Workers

For someone mid-career, the hiring data cuts two ways. Your seat looks safer than the headlines suggest. But fewer juniors means more of the routine work lands on you or on a tool, and the path you climbed may not exist for the next person.

Which Jobs AI Touches Least

Jobs AI barely touches share at least one of three traits. They happen in a physical place that changes from day to day. They depend on trust between two people. Or they carry responsibility that someone has to put their name to.

What kind of work stays human for now

• Hands in an unpredictable space. An electrician rewiring a 1950s house, a nurse turning a patient, a cook on a packed Friday night. Each job differs from the last, and the body is doing the work.
• Trust built over time. Childcare, counseling, selling to people who buy from someone they know. The relationship is the product.
• Accountability. Someone has to approve the bridge design, sign the audit, or choose the treatment. AI can lay out the options, but a person carries the call.

Pew also found that the most exposed jobs pay more: about $33 an hour on average, against $20 in the least exposed. College graduates were more than twice as likely as high school graduates to hold a highly exposed job, 27% versus 12%. This time, automation is aimed at the better-paid, more educated office worker.

For a role-by-role list, our guide to careers safe from AI ranks specific jobs. The point here is narrower: "safe" describes a kind of work, and some slice of almost every job falls into it.

New Jobs AI Is Creating as It Changes Old Ones

AI adds jobs as well as removing them.

Where the new roles are actually showing up

Employers in the World Economic Forum's survey expect 170 million new jobs by 2030 and 92 million lost ones, which nets out at 78 million more. They cover every force reshaping work, not AI alone, and they are employer forecasts, not promises.

The new roles fall into three groups:

• Building it. AI and machine learning specialists, big data specialists, and software developers lead the fastest-growing list.
• Feeding and checking it. Data labelers, trainers who write and grade examples for AI models, and reviewers who catch the model's mistakes. Much of this is contract or gig work, paid by the task.
• Using it inside an old job. The biggest group, and the hardest to see. A marketer who runs AI tools for a small team, a nurse who manages an AI note-taker, an accountant who reviews machine-prepared returns. Nothing on the job board says "new," but the job is new.

The new jobs rarely go to the people who lost the old ones. A data entry clerk in Ohio does not become a machine learning engineer in San Francisco. That mismatch in skills, place, and pay is where the cost of this change lands.

How AI Is Changing the Skills Jobs Require

Even jobs that survive whole are asking for different things. The Forum's employers expect 39% of the core skills workers use today to change by 2030. That is the less visible side of how AI is changing jobs: the title stays while the skills behind it shift.

The skills rising fastest inside existing jobs

Employers put AI and big data skills at the top of the fast-rising list, followed by networks and cybersecurity, and general comfort with technology. Human skills climb right behind them: creative thinking, resilience and flexibility, and curiosity with a habit of learning.

Inside a role, the shift tends to look like this:

• From producing a first draft to judging one. The writer edits more than they write. The analyst checks the model's numbers instead of building every sheet from scratch.
• From knowing the answer to framing the question. Asking the tool the right thing, with the right background, becomes part of the job.
• From doing the task to owning the result. When the tool gets it wrong, the person who signed off still answers for it.
• More time with people. As the screen work shrinks, more hours land on clients, teammates, and decisions.

The skill that grows most is judgment: knowing when the machine's answer is good enough. It is hard to learn from a course and easier to learn from years in a field, which helps explain why experienced workers have held up better in the Stanford data.

Learning the tools still matters. If you want to use AI on your own career questions, the AI career planning guide walks through getting useful answers from a chatbot and checking them.

How to Know If AI Is Changing Your Job Specifically

Industry averages describe a crowd. You need to know about one week: yours. It takes about half an hour with a list and a pen.

A quick way to read your own exposure

1. List last week's tasks. Skip the job description and write what you did, with rough hours: "wrote three client updates, 4 hours," "matched invoices, 6 hours," "ran the Tuesday team meeting, 1 hour."
2. Put each task in one of three groups. Group A: a current AI tool could produce a usable first version today. Group B: AI could help, but the result needs your judgment or your knowledge of the situation. Group C: it needs your hands, your presence in the room, or someone's trust in you.
3. Add up the hours. Work out what share of your week sits in Group A.
4. Read what is left. Treat the B and C tasks as a job description on their own. That is roughly the job you will have in a few years if you stay.

As a rough guide: if under a quarter of your hours fall in Group A, AI will mostly change your tools. From a quarter to a half, it will change your role. Above half, expect the role to be redesigned, merged with another, or hired for differently.

The question that matters is whether you want the job made of what is left. Your answer decides your next move.

Signs your role is already shifting under you

• Your team has stopped replacing junior people who leave.
• Your manager wants the same output in less time and points you to a tool for it.
• Job ads for your role at other companies now list AI tools under requirements.
• A task you used to own now arrives already done, and your part is to check it.

Two or more of these and the change is under way, whatever your company has announced.

What To Do When AI Is Changing Your Job

Most advice on how to prepare for AI changing your job stops at "learn new skills." That helps, but it skips the decision underneath. Once you know what share of your week is exposed, you have three moves, and each calls for different preparation.

How to prepare without guessing

Match the move to what your task sort showed:

Whatever the move, a few steps pay off now. Use AI tools on your own work for a month so you know their limits firsthand. And build a small savings cushion: the Stanford data shows companies adjusting through hiring, so your next job search may take longer than your last.

If the worry itself is crowding out clear thinking, our guide to existential AI career advice deals with that side of it.

When the right move is a different path, not a new skill

Sometimes the sort shows that AI is taking the part of the job you loved and leaving you the part you put up with. A designer who loved making things ends up reviewing machine-made drafts all day. No course solves that, because the job itself has turned into a different job.

If AI takes the part of your work you enjoyed, a new skill will not bring it back. That is the signal to look at a different path, and it is a better reason to move than fear.

The hard part is knowing what to move toward. Most people pick their next role by what they are qualified for, which is often how they landed in the last one. A better starting point is the work you do best and want to keep doing, plus the setting you need around it. The Pigment career test can give you that picture. It costs $79 and shows which work you are strongest at, what you need around that work to do it well for years, and the directions worth testing next. Most people who take it are mid-career, a decade or more in, facing this kind of turn.

Once you know where you are headed, the best career change tools roundup covers the practical side.

FAQ: How Is AI Changing Jobs?