The Anthropic Economic Index: How AI Is Actually Used

The Anthropic Economic Index is a research program that measures how AI is actually used in the economy, built from anonymized Claude usage data. Launched in 2025, it provides a data-driven picture of which tasks, occupations, and industries are adopting AI — information that debates about AI's labor market impact often lack. This article explains how the index works, what it has found, and why it matters for policymakers, businesses, and workers.

Background

• The index launched in February 2025 as a quarterly research product, using anonymized and aggregated data from Claude conversations to analyze how people use AI in real work. The methodology is careful about privacy: data is aggregated, filtered, and analyzed without identifying individual users.
• The core concept is task-level analysis: instead of asking whether AI replaces jobs, the index examines which specific tasks within occupations involve AI, revealing partial automation — AI used alongside human work — as the dominant pattern.
• Its findings have consistently shown that AI use is concentrated in software development, writing, and analytical tasks, and that AI is used more to augment work than to fully automate it — a nuance that flat "job replacement" narratives miss.

Key facts

Item	Detail
Launched	February 2025
Data source	Anonymized Claude usage
Unit of analysis	Tasks and occupations
Key finding	Partial automation dominates
Top use areas	Coding, writing, analysis
Publication	Quarterly reports
Privacy	Aggregated, anonymized
Audience	Policy, business, research

Highlights

How the index works

The methodology turns messy usage data into economic insight: conversations are classified by task and matched to occupational categories, then aggregated into shares of usage. The result is a map of where AI is genuinely deployed — not where it could be. The image below evokes the data infrastructure behind the measurement:

Caption: The Economic Index converts anonymized usage data into task-level economic insight — a rare, real-world measure of AI adoption.

What the data shows

The consistent findings: AI usage clusters in software engineering, writing and editing, and analytical work; most usage involves AI augmenting tasks rather than replacing whole jobs; and adoption is broad across skill levels. The nuance — partial automation — is the index's most valuable contribution.

Why it matters

For policymakers, the index provides evidence for regulation and labor market planning; for businesses, it reveals where AI adoption is actually happening; for workers, it shows which tasks are being augmented. It is a data anchor in debates that often run on anecdotes.

Industry positioning & impact

The Economic Index is a genuinely distinctive asset: Anthropic is the only frontier lab publishing systematic, task-level measurement of real AI usage. The strategic value is multi-layered. For the policy debate, the index supplies evidence on partial automation and usage concentration that shapes how governments think about AI and labor. For the business conversation, it tells enterprises where competitors are actually deploying AI, which informs adoption strategy. For Anthropic itself, the index builds authority: a company that measures the economy's AI usage positions itself as the responsible, data-driven leader in a field of hype. The competitive implication is that other labs now face a choice — publish comparable data or concede the measurement position. The index also feeds Anthropic's safety narrative: understanding real usage is part of responsible deployment. As of 2026, the open questions are how the index evolves methodologically, what new findings emerge as adoption broadens, and whether it becomes the standard reference for AI labor economics. The index reports on anthropic.com are the authoritative source.

Related reading

For the research program behind the index, see Anthropic Research: Safety, Interpretability, and AI; for the models generating the data, Anthropic Claude Models: Haiku, Sonnet, and Opus; and for the company context, What Is Anthropic AI? Company, Mission, and Models.

References

The authoritative source is the Anthropic Economic Index on anthropic.com, including its methodology and quarterly reports. For broader labor market context, the US Bureau of Labor Statistics provides official occupational data.

Buying advice & audience

If you are searching "anthropic economic index", "how is ai actually used", or "ai labor market data 2026", here is the practical guidance. For policymakers and analysts, the index reports are primary sources — cite the methodology and the partial-automation finding rather than secondary commentary. For business leaders, the task-level data is adoption intelligence: see where usage concentrates and compare it with your own AI deployment plans. For workers and students, the index is a realistic picture of how AI augments work — useful for career planning that assumes augmentation, not replacement. For researchers, the methodology is a template for measuring AI adoption, and the quarterly cadence provides an ongoing dataset. For anyone reading media coverage of the index, check the original reports for the nuance — the "augmentation over replacement" finding is frequently flattened in headlines. The related articles cover the research agenda and the models; this guide gives you the index's meaning and use.

FAQ

What is the Anthropic Economic Index?

It is a research program that measures how AI is used in the economy by analyzing anonymized, aggregated Claude usage data at the task level. Launched in 2025, it publishes quarterly reports on which tasks, occupations, and industries are adopting AI.

How does the index measure AI usage?

The methodology classifies Claude conversations by task, matches tasks to occupational categories, and aggregates the results into usage shares. Data is anonymized and aggregated to protect privacy, and reports focus on patterns rather than individuals.

What has the index found?

The consistent findings: AI usage concentrates in software development, writing, and analytical work; partial automation — AI augmenting tasks within jobs — is the dominant pattern rather than full replacement; and adoption is broad across skill levels. The nuance matters for labor market debates.

Why does the index matter?

It provides real-world evidence in debates that often run on anecdotes: policymakers get data for labor planning, businesses get adoption intelligence, and workers get a realistic picture of augmentation. It is a reference point for AI's economic impact.

Where can I read the Anthropic Economic Index?

The reports and methodology are published on the Anthropic Economic Index page at anthropic.com/economic-index. It is the authoritative source, with each quarterly release adding new findings on how AI is actually used.