Short answer: some jobs, yes, and all jobs, no. The studies we read say AI changes the tasks inside jobs more than it deletes the jobs. One more piece of evidence is on our own list: on 26 October 2026, 2,544 open roles on 23 platforms ask people to write, check and correct what AI produces, and many of them ask for doctors, lawyers, programmers and translators.

Watch on YouTube, 8:34Will AI Replace Your Job? What 2,000 AI Listings Reveal

Cover: frame from The Day After AGI, World Economic Forum, CC BY 4.0

Prefer to watch? The 8 minute video Will AI Replace Your Job? What 2,000 AI Listings Reveal goes through the same numbers on YouTube.

What the big studies say

  • US Bureau of Labor Statistics: in its projections to 2035, published on 27 August 2026, total employment grows 3.5%. The group that shrinks most is office and administrative support, at minus 4.0%, and the BLS names "automation tools, including those powered by AI" as a likely reason.
  • IMF: in January 2024 it said almost 40% of jobs worldwide are exposed to AI, and about 60% in advanced economies. Exposed does not mean replaced: roughly half of the exposed jobs may gain productivity, and the other half may see lower demand, wages and hiring.
  • Goldman Sachs Research: in August 2025 its baseline was that 6% to 7% of US workers could be displaced if AI is widely adopted, with the effect expected to be temporary. In a July 2026 episode of its Exchanges podcast, its economist Joseph Briggs gave a newer figure: around 9% of US workers reallocated over ten years.

These are projections made at a point in time, and each source warns about uncertainty. Read them as a direction, not a forecast for your own job.

Tasks go first, not whole jobs

The BLS ties the fall in office support to automation tools, which points at routine tasks rather than at whole occupations. The IMF splits its exposed jobs in two: some gain, some lose. So the useful question is not "will AI replace my job" but "which of my tasks are routine, and which need judgment".

What AI still asks people for

AI learns from people. Someone has to write expert answers, check the reasoning and flag the mistakes, and that work is posted as paid roles. We read the public job lists of 23 platforms every day. On 26 October 2026 there were 2,544 open roles, and 1,408 of them listed an hourly rate, with a median of $55.

The more specialised the field, the higher the listed rate:

FieldHourly ratesMedianTop 10% start at
Law108$110$270
Medicine106$100$200
Finance86$90.50$160
Coding172$70$125
STEM220$69$105
Writing59$67.50$100
Trades7$60-
Generalist211$35$100
Languages427$22.50$52.50
Robotics12$10.50-

Medicine has 140 open roles and a median listed rate of $100. Law has 121 and $110. Languages have the most roles, 1,150, but only 427 list a rate, and the median there is $22.50.

Every figure is a rate that a platform lists. It is the pay in the ad, not what people earn, and a role is not a promise of work: platforms decide who they accept, and projects come and go.

2,605 open roles at 22 platforms today

See all 2,605 rolesCompare the platforms

Best-paid expert roles

For doctors, lawyers, PhDs and scientists: the highest hourly pay listed on the 23 platforms we read. Chosen by pay only, at most two per platform.

Referral link: we may earn a fee if you join through it. How this works

What to do with this

  1. Look at your tasks, not your title. The routine, repeatable ones are the most exposed.
  2. Look at the parts that need judgment and deep knowledge. Those are the parts these roles still ask people for.
  3. Learn to work with the tools. Checking and correcting AI output is itself a skill that employers list.

For a view by profession, with the listed rates for each field, see which professions these roles ask for, one page each. The full table of listed rates is on the pay index.

Prepare the interview: Doctor, Nurse and Health Professional interview questions.

Sources

New AI training roles by email, once a week. No password.

Get alertsAll roles