Skip to sources
Time Machine

Secretaries and Administrative Assistants

Scrub through 162 years of this role's history — from when it first emerged, through every wave of technology that reshaped it, to the cited projections for where it's heading next.

Mechanical typewriterMechanical typewriter
IBM MT/STIBM MT/ST
Microsoft Word + WordPerfect on PC
Microsoft Outlook + corporate email
Google Calendar + cloud-shared scheduling
AI scheduling assistants (Clara, x.ai Amy)
LLMs in-line in the office suite (MS 365 Copilot, Gemini for Workspace)LLMs in-line in the office suite (MS 365 Copilot, Gemini for Workspace)
DictaphoneDictaphone
Xerox 914 photocopierXerox 914 photocopier
Wang Word Processing SystemWPWang Word Processing System
187519001925195019752000now

Drag the dot, click anywhere on the track, or use ← → arrow keys (Shift for 10-year jumps, PgUp/PgDn for 25).

2026
Known today as Administrative Assistant
US Employment
1.94M
BLS OOH headline for 43-6014. Note: BLS bundles all 43-60xx in the OOH headline as "3.5M jobs" — the 1.94M figure here is the General category alone.
Median Annual Wage
$46,290
Source: BLS-OEWS
LLMs in-line in the office suite (MS 365 Copilot, Gemini for Workspace)Tool of the era · LLMs in-line in the office suite (MS 365 Copilot, Gemini for Workspace)

Generative AI embedded directly in Word, Outlook, Excel, Teams (Copilot, GA Nov 2023, $30/user/mo) and Google Docs/Sheets/Gmail (Gemini, broadly available 2024, $14/user/mo). Drafts replies, summarizes threads, generates documents — direct overlap with the drafting + summary loop that defines admin work.

By Nov 2025 Office & Admin Support tasks were 13% of all Claude API transcripts (up from 10% in Aug 2025), per Anthropic Economic Index — the first observational evidence of real substitution at scale.

Projection cone · present → 2034

What credible sources project

Scrub the slider past now to anchor each scenario on the scrubber. The spread you see below is the range of futures credible sources project for this role.

BLS Occupational Outlook 2024
2034
-1%
BLS Employment Projections — industry-occupation matrix + labor productivity assumptions. Most conservative source: assumes productivity gains absorb displacement rather than triggering layoffs.
Anthropic Economic Index (live observational)
2026
-13%
Direct measurement of Claude API usage by task category (January 2026 report). Office & Admin Support tasks = 13% of all transcripts in Nov 2025, up from 10% in Aug 2025. Only data source observing actual substitution in production. The 13% figure is *current usage share*, not a forecast — but it directly maps to "AI is doing some real fraction of this work today". baselineYear re-anchored to the 2024 OEWS employment anchor (the most recent published anchor in this profile's employment series) so the scrubber fan-out renders cleanly.
McKinsey Global Institute
2030
-25%
Generative AI exposure × adoption-curve modeling. Specifically projects 710,000 administrative assistant jobs lost by 2030 in midpoint scenario; office support category as a whole -18%.
Goldman Sachs (March 2023)
2030
-46%
Goldman maps O*NET work-activity importance scores to LLM capability ratings (March 2023 report). Result is share of *tasks automatable*, not jobs lost — but 46% is the highest task-automation share among any US occupation category. Interpret as ceiling, not floor. baselineYear re-anchored from the report's 2023 task profile to 2022 OEWS employment (the nearest published anchor in this profile's employment series) so the scrubber fan-out renders cleanly.
Eloundou et al. — "GPTs are GPTs" (2023, observational)
2024
-64%
GPT-4 task-by-task labeling against O*NET task statements (published March 2023). For 43-6014: α (fully exposed) = 0.385, β (E1 + 0.5×E2) = 0.644, γ (any exposure) = 0.904. Reported here as β = -64% to fit the cone display; γ would read as -90%. baselineYear re-anchored to the 2022 OEWS employment anchor (the nearest published anchor in this profile's employment series prior to the paper's release) so the scrubber fan-out renders cleanly.
Frey & Osborne (2013)
2033
-96%
Gaussian-process classifier on O*NET task features; 70 expert-labeled occupations as training set. Reports "probability of computerisation" ~0.96 for 43-6014. Exact appendix decimal needs human verification of the PDF table — value used here is conventionally cited. F&O's analysis was published in 2013; baselineYear re-anchored to the 2017 OEWS employment anchor (the earliest post-publication year in this profile's employment series) so the scrubber fan-out renders cleanly. The implied 20-year horizon is unchanged.
Today, in this role

What's shifting in the work right now

The historical view above shows how this role has moved. This is the present-day detail: which AI tools are picking up which tasks, where the edge still is, and the natural directions this work can grow.

What's changing in your day

Three parts of your work where AI is already doing real lifting — and what stays yours.

AI is sitting alongside you here

Review work done by others to check for correct spelling and grammar, ensure that company format policies are followed, and recommend revisions.[2]

Where your edge is

AI is sitting alongside you here

Order and dispense supplies.[2]

Where your edge is

AI is sitting alongside you here

Answer telephones and give information to callers, take messages, or transfer calls to appropriate individuals.[2]

Where your edge is

Present-day sources

Sources

Every claim on this page traces back to one of the following. Updated 2026-05-30.

  1. [1]Eloundou et al. 2024 — GPTs are GPTs (Science)· accessed 2026-05-30
  2. [2]O*NET 30.3 — US Department of Labor· accessed 2026-05-30
Share this year
Drops anyone you send it to straight into 2026.
Preview card
Different role?

See the same long-arc view for your own profession.

Browse the directory by industry, or search by title or SOC code. New roles ship every few weeks — every profile cites every claim.

Browse all roles