Anthropic's AI Job Impact Metric: Programmers 75%

🔑 Enhanced Key Takeaways

• •Anthropic's index identifies customer service representatives, data entry clerks, and medical record specialists as highly exposed professions alongside programmers, with 30% of jobs showing no exposure due to human-centric nature.
• •AI exposure correlates strongly with real-world Claude usage, with ICT professionals leading adoption, followed by gradual spread to other sectors, particularly in high-income countries.
• •Adjusting for task success rates and time weighting revises productivity growth estimates downward from 1.8 to 1.0-1.2 percentage points annually over the next decade.
• •Suggestive evidence shows decelerated hiring for young workers aged 22-25 in vulnerable occupations, despite no overall unemployment rise.
• •Nearly 49% of occupations now show Claude usage for at least a quarter of tasks when pooling data across reports.

🛠️ Technical Deep Dive

• •Measured Exposure combines job function analysis, LLM capability approximations, and anonymized Claude usage data from November 2025 conversations, categorized by task, use case (professional/educational/personal), and AI autonomy levels (collaboration to full delegation).
• •Incorporates task success rates, duration weighting, and time share of tasks within occupations to compute effective exposure, revealing higher impacts for roles like data entry keyers and radiologists.
• •Distinguishes automation (task removal leading to potential deskilling) from augmentation, with rising 'directive' use indicating increasing delegation.

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline