JOB FORECAST

HOW MUCH OF YOUR JOB'S WORK DOES AI TOUCH?

Pick your job to see how much of its work already shows up in how people use AI at work, and which tasks. 756 jobs. It measures exposure, not job loss.

THE MOST EXPOSED JOBS

  1. Computer Programmers 75%
  2. Customer Service Representatives 70%
  3. Data Entry Keyers 67%
  4. Medical Records Specialists 67%
  5. Market Research Analysts and Marketing Specialists 65%
  6. Medical Transcriptionists 64%
  7. Sales Representatives, Wholesale and Manufacturing, Except Technical and Scientific Products 63%
  8. Database Architects 58%
  9. Financial and Investment Analysts 57%
  10. Software Quality Assurance Analysts and Testers 52%

At the other end, 54% of the jobs have none of their work showing up often enough yet.

ALL JOBS, A TO Z

A (56)
B (27)
C (93)
D (26)
E (47)
F (68)
G (20)
H (35)
I (19)
J (4)
K (1)
L (29)
M (63)
N (14)
O (22)
P (81)
R (32)
S (62)
T (36)
U (4)
V (3)
W (13)
Z (1)

HOW IT'S MEASURED

Observed AI exposure comes from Anthropic's researchers. They took every task of each job from the U.S. government's O*NET job database, kept the tasks a 2023 study found AI could at least double the speed of, and checked which of those show up in people's work use of Claude. Each task is weighted by how much of the job's time it takes. Tasks AI does on its own count fully; tasks where it helps a person count half. The result is the share of the job's working time that's "exposed" in practice.

Where AI use shows up lists the job's tasks that appear in that usage, the most common first. A task can appear here even when the job's exposure is 0%: it may take little of the job's time, or fall outside what the 2023 study counted as speedable.

Not showing up yet lists the job's most important tasks, as rated by people in the job for O*NET, that don't appear in the usage at all.

The researchers found "no systematic increase in unemployment for highly exposed workers since late 2022", and only tentative signs of slower hiring of young workers in exposed jobs. Read their study

What we changed in the O*NET data

Tasks from an occupation's detailed O*NET codes are merged under its six-digit code, to match the exposure data. Tasks are ranked by O*NET's importance rating, and we show at most eight tasks where AI use shows up and five where it doesn't. Task wording is unchanged.

Exposure and task usage: Anthropic Economic Index, "Labor market impacts of AI" (Massenkoff and McCrory, 5 Mar 2026), data under CC BY. It measures Claude usage only.

This page includes information from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. AI Weather Report has modified all or some of this information (what we changed). USDOL/ETA has not approved, endorsed, or tested these modifications.