Insights · run 1.0.0-draft.1 · 2026-08-05

What the index says, in aggregate

Employment- and wage-weighted headline metrics across 887 scored occupations covering 194.2M U.S. workers (BLS OEWS wages and employment). Pressure, not prophecy — see how to read this below.

Share of the U.S. wage bill under substitution pressure

32%

$4.70T of $14.47T in annual median wages, substitution-score-weighted across occupations. This measures how much paid work sits in exposed tasks — not wages that will disappear.

Employment-weighted substitution

34

Average substitution score experienced by the typical U.S. worker (0–100).

Employment-weighted augmentation

55

The counterpoint: how much AI assists the typical worker's tasks without replacing them.

Workers in high-pressure occupations

5.3M

Employed in occupations with substitution ≥ 70 — 2.7% of covered employment.

Workers in high-augmentation occupations

47.8M

Employed in occupations with augmentation ≥ 70 — 24.6% of covered employment.

Highly exposed tasks

7.5%

Of 18,796 rated O*NET tasks, the share scoring ≥ 70 on substitution.

Average confidence-interval width

±6.7

Mean half-width of the substitution CI from rater disagreement — the index's own uncertainty, stated plainly.

Substitution × augmentation — every occupation

The core claim of the methodology in one picture: replacement and assistance are different axes. Each dot is one of 923 occupations; dot area reflects U.S. employment. Hover a dot for its name.

00252550507575100100AI assists, rarely replacesdeep task transformationlittle AI contact todayreplacement pressureSubstitution →Augmentation →

The shape of the index

Distribution of all 923 occupation substitution scores; the marker is the employment-weighted mean.

Pressure by wage decile

Employment-weighted mean substitution per median-wage decile — where on the pay scale the pressure sits.

$31k–$40k$40k–$46k$47k–$50k$50k–$59k$59k–$65k$66k–$76k$77k–$85k$86k–$102k$102k–$126k$126k–$559k

By occupation group

Employment-weighted mean substitution per SOC major group.

Top 20 by substitution

full index →
Top 20 occupations by substitution score
#OccupationSubstitution (CI)ExposureAugment.Median wageEmployed
1Word Processors and Typists89.2(8692)8967$49,28035,010
2Travel Agents83.1(7394)8172$50,16055,110
3Bookkeeping, Accounting, and Auditing Clerks81.4(7687)8478$50,6701,373,680
4Telemarketers80.8(7686)8068$35,45058,430
5Order Clerks78.0(7284)7471$46,17075,200
6Payroll and Timekeeping Clerks77.2(7085)7971$58,260153,140
7Billing and Posting Clerks75.8(6785)7666$48,500404,060
8Insurance Claims and Policy Processing Clerks75.5(6982)7675$49,230214,260
9Brokerage Clerks75.1(6882)7771$65,75035,940
10Data Entry Keyers74.4(7079)7255$41,340127,080
11Correspondence Clerks72.8(6581)7274$46,8004,290
12Statistical Assistants71.6(6678)7078$50,3304,710
13Online Merchants71.2(6479)6775$83,0501,087,090
14Secretaries and Administrative Assistants, Except Legal, Medical, and Executive70.3(6378)6778$47,5401,706,790
15Proofreaders and Copy Markers70.1(6476)6582$51,1204,580
16Switchboard Operators, Including Answering Service69.5(5980)6754$38,63034,280
17Tellers68.5(6077)7063$43,030329,480
18Loan Interviewers and Clerks68.5(6077)7076$50,020164,790
19Credit Authorizers, Checkers, and Clerks68.0(6274)6866$50,08012,030
20Receptionists and Information Clerks67.9(6274)6569$38,010910,180

How to read these numbers

"Wage bill under pressure" weights every occupation's wages by its substitution score — it measures where paid work overlaps with what AI can plausibly take over, not a payroll forecast. Realized effects so far are concentrated and cohort-specific, and augmentation-weighted usage still dominates. Metrics cover the 887 scored occupations with BLS wage and employment data; methodology and formulas are public on the methodology page.