Anthropic Economic Index — How Claude Is Actually Used Across Industries

Unlike most AI labor market studies, which model hypothetical futures, the Anthropic Economic Index measures what is actually happening right now. Published in March 2025, it analyzed millions of real conversations with Claude to map which occupations and industries are using AI, how they are using it, and whether usage patterns look more like automation (replacing human tasks) or augmentation (enhancing human capability).

Key Findings

• Software development and writing dominate AI usage. Computer and mathematical occupations account for the largest share of AI interactions, followed by business, arts/media, and sales roles. Software developers are by far the heaviest users of AI in their daily work.
• Most usage is augmentation, not automation. Across all occupations, the study found that roughly 57% of AI interactions are augmentative — humans using AI to enhance their work — while about 43% involve more direct automation of tasks. Only a small fraction represents full task displacement.
• Mid-skill occupations use AI more than expected. Contrary to the assumption that AI primarily serves elite knowledge workers, the data shows significant adoption in customer service, sales, administrative support, and healthcare support roles. AI use is not confined to tech workers.
• Usage does not match expert predictions. The occupations that experts predicted would be most affected by AI do not perfectly overlap with the occupations that are actually using AI the most. Some high-exposure occupations have low adoption, while some "safe" occupations show heavy usage.
• Education and healthcare show growing adoption. Teaching and healthcare support roles are among the fastest-growing categories of AI usage, primarily in augmentation mode — preparation, documentation, and information synthesis rather than direct patient or student interaction.

What This Means for Your Career

The Anthropic Economic Index provides a reality check against forecasts. If you have been reading studies that predict your job will be automated, this data shows that prediction and reality often diverge. The occupations using AI most heavily are not necessarily the ones being replaced — they are the ones where workers have found practical ways to integrate AI into their workflows.

The augmentation-dominant pattern is particularly important. If you are in a role where AI could help with documentation, drafting, analysis, or information retrieval, the data suggests your peers are already using it for exactly those tasks. Not adopting AI in these contexts puts you behind, not ahead.

The finding about mid-skill occupations should reassure some workers and concern others. AI is not just a tool for software engineers and executives. Customer service representatives, sales professionals, and administrative workers are actively using AI — which means AI literacy is becoming table stakes across skill levels, not just at the top.

Data Highlights

• 57% of AI interactions are augmentative (enhancing human work)
• 43% involve more direct task automation
• Software development is the single largest occupational category of AI usage
• 4x growth in education-sector AI usage over the study period
• 37% of occupations with high predicted AI exposure have below-average actual adoption

Methodology

The Anthropic Economic Index analyzed a large-scale sample of privacy-preserving conversation data from Claude usage, classifying interactions by occupation (using the U.S. Bureau of Labor Statistics Standard Occupational Classification system) and by task type. Each interaction was scored on an automation-augmentation spectrum. The research team cross-referenced usage patterns against existing AI exposure frameworks (including the OECD and ILO models) to identify gaps between predicted and actual AI adoption. The analysis covers millions of interactions across hundreds of occupational categories.