12X Faster, But Your Job at Risk: Anthropic's $350B AI Warning
Anthropic's report: AI speeds complex tasks 12x but may deskill jobs. Insights from $350B valuation firm.
“AI Disruption” Publication 8600 Subscriptions 20% Discount Offer Link.
Is AI Really Making People Work Faster? What Tasks Does It Support Best? And How Will It Transform the Nature of Work?
Anthropic is attempting to answer these core questions with real-world data through its ongoing “Economic Index” project. By conducting privacy-protected analysis of conversations on Claude.ai (primarily for consumer use) and its first-party API (primarily for enterprise use), Anthropic gains insights into how AI is actually being used.
In its latest fourth report, Anthropic introduces a new set of analytical tools called economic primitives. These five simple and fundamental metrics are designed to track Claude’s economic impact over the long term. They are:
Task complexity
Skill level
Purpose (work, education, or personal)
AI autonomy
Success
This latest report, based on conversation samples from November 2025 (primarily using the Claude Sonnet 4.5 model), uses these “primitives” to explore a range of complex questions, such as: How does Claude’s success rate change as tasks become more complex? And do current AI usage patterns suggest that many jobs will face a “deskilling” effect?
Here’s a summary of the report’s key findings:
What Do the Economic Primitives Reveal?
The report applies the economic primitives at three levels: tasks, occupations, and overall impact.
