Will AI Take My Job? What the Research Actually Says

The short answer

AI takes over tasks faster than it takes over whole jobs. Most jobs are a bundle of tasks, and AI is good at some of them, like reading, writing, summarizing, and answering routine questions, and bad at others, like physical work, judgment calls, and building trust. The jobs most at risk are the ones where the AI-friendly tasks make up most of the day. The clearest effect so far is on young workers starting out in those fields.

What the research says

Read together, they point the same way: a lot of work is exposed, the net effect on total jobs is unclear, and the early damage is concentrated at the entry level.

Jobs most exposed to AI

From Microsoft's study, the jobs whose tasks overlap most with what AI already does:

1. Interpreters and translators
2. Historians
3. Passenger attendants
4. Sales representatives of services
5. Writers and authors
6. Customer service representatives
7. CNC tool programmers
8. Telephone operators
9. Ticket agents and travel clerks
10. Broadcast announcers and radio DJs

The pattern: knowledge work done at a computer, especially work that's mostly communicating information.

Jobs least exposed to AI

From the same study, the jobs with the least overlap:

1. Dredge operators
2. Bridge and lock tenders
3. Water treatment plant operators
4. Foundry mold and coremakers
5. Rail-track laying equipment operators
6. Pile driver operators
7. Floor sanders and finishers
8. Orderlies
9. Motorboat operators
10. Logging equipment operators

The pattern: hands-on, physical, or in-person work. The World Economic Forum also expects growth in delivery drivers, care workers, educators, and farmworkers.

Will AI replace lawyers?

AI is strong at legal research, document review, and first drafts, which make up much of junior legal work. It can't take responsibility for advice, negotiate, appear in court, or be accountable when things go wrong. Expect fewer hours spent on routine work and pressure on junior roles, not the end of lawyers.

Will AI replace accountants?

The routine parts, like data entry, categorizing transactions, and reconciliations, are already being automated. Judgment, advisory work, tax planning, and signing off on the numbers still need a person who is responsible for them. Accountants who hand the routine work to AI are better placed than those who don't.

Will AI replace programmers?

I can speak to this one from experience. I build software with AI coding agents every day, and the 2,400+ tools on this site were built with them. They write a lot of code fast, but they also make confident mistakes that someone has to catch, which I wrote about in what AI coding agents are still bad at. The work is shifting from typing code to deciding what to build and checking what the AI produced. That hits entry-level developers hardest, which matches the Stanford data.

How to judge your own job

Ask yourself these questions:

• How much of my day is reading, writing, or answering questions on a computer? The higher the share, the more exposed you are.
• How much depends on judgment calls I'm accountable for? Responsibility is hard to hand to software.
• How much depends on relationships and trust? People still want to deal with people for things that matter.
• How much is physical or in person? That remains the hardest part for AI.

What to do about it

• Learn to use AI on your own tasks. The people most at risk are the ones competing with AI instead of using it.
• Move toward the parts AI can't do. Judgment, relationships, and hands-on skill.
• Watch the entry level. If you're early in a career in an exposed field, build skills that AI makes more valuable, not less.

For businesses, the same picture shows up in the numbers. An AI agent can take on part of a role, never all of it, which is how the AI ROI calculator is built. And if you want the bigger picture, what happens to an economy when machines do most of the work is the question behind The Post-Labor Economy, a thought experiment in the Lab.