AI and Education: What 152,000 Conversations Reveal

Abstract

The Anthropic Economic Index V4 sampled one million Claude.ai conversations during a single week in November 2025 and classified each against standardised occupational tasks. This report examines the 152,088 conversations (15.2% of total conversation volume) that mapped to education occupations.

Three patterns stand out. First, students are the primary users: 59.5% of education-related Claude.ai usage is classified as coursework, compared with 18.9% across all tasks. Second, interaction styles are dominated by directive requests (34.6%) and task iteration (33.0%), while feedback loops are nearly absent (1.6% vs 13.8% for all tasks). Third, education subsectors show distinct patterns: K-12 teachers have the highest learning-classified interactions (36.9%), while postsecondary usage is concentrated in task iteration (34.0%).

These findings describe how people used one AI platform during one week. They do not measure learning outcomes, generalise to other platforms, or establish whether interaction patterns reflect pedagogical intent. The data source, scope, and limitations are described in full.

This is Report 3 in the AlignED series. For all reports, visit AlignED Reports.