# Anthropic ‘Observed Exposure’: Augmentation So Far, Not Mass Job Kill

Times of AI Desk · 2026-03-05 · Industry

[https://timesof.ai/2026/03/anthropic-labor-market-impacts-ai-report](https://timesof.ai/2026/03/anthropic-labor-market-impacts-ai-report)

> Anthropic Economic Research (March 5): ‘observed exposure’ blends LLM capability with real Claude usage — highest in digital white-collar (programming ~75%, customer service ~70%), near-floor in physical work. Adoption still below theoretical potential; no unemployment differential spike since late 2022. Lab usage data — not BLS causality.

Task-exposure papers usually stop at “what an LLM *could* do.” Anthropic’s proprietary metric: **observed exposure** — capability **weighted by actual Claude usage** — so the labor story is adoption practice, not only theoretical reach.

**Anthropic Economic Research** released “Labor market impacts of AI: A new measure and early evidence” (March 5; Maxim Massenkoff and Peter McCrory). Findings: highest observed exposure in office digital tasks — programming (~**75%** / computer programmers **74.5%**), data entry, customer service (**~70.1%**); very low in physical/hands-on roles (construction, agriculture, machinery). AI primarily **augmenting** rather than fully automating; tentative signals of slower entry for young workers (**22–25**) in exposed occupations and wage premiums in highly exposed roles; **no detectable rise** in unemployment differentials between exposed and insulated jobs since late **2022**. Current models still fall short of capability ceiling in practice. Draws on U.S. labor data + Anthropic usage patterns.

## Claims vs checks

Method and figures are **Anthropic research primary**. Usage is Claude-centric — not all AI tools. Fortune/LinkedIn coverage is secondary. Absence of unemployment spike is an early-window observation, not a forecast.

## Limits

- One lab’s traffic ≠ economy-wide AI mix.
- Augmentation today can become substitution later.
- Young-worker entry signals are tentative as labeled.

## Sources

- [Anthropic: “Labor market impacts of AI: A new measure and early evidence”](https://anthropic.com/research) (March 5, 2026).
- Key findings on observed exposure and occupation coverage rates.
- Fortune, LinkedIn analyses, and economic roundups (March 2026).