Anthropic Education Report How University Students Use Claude

Anthropic Education Report: How University Students Use Claude \ Anthropic Announcements Societal Impacts Anthropic Education Report: How university students use Claude Apr 8, 2025

AI systems are no longer just specialized research tools: they’re everyday academic companions. As AIs integrate more deeply into educational environments, we need to consider important questions about learning, assessment, and skill development. Until now, most discussions have relied on surveys and controlled experiments rather than direct evidence of how students naturally integrate AI into their academic work in real settings. To address this gap, we’ve conducted one of the first large-scale studies of real-world AI usage patterns in higher education, analyzing one million anonymized student conversations on Claude.ai. The key findings from our Education Report are: STEM students are early adopters of AI tools like Claude, with Computer Science students particularly overrepresented (accounting for 36.8% of students’ conversations while comprising only 5.4% of U.S. degrees). In contrast, Business, Health, and Humanities students show lower adoption rates relative to their enrollment numbers. We identified four patterns by which students interact with AI, each of which were present in our data at approximately equal rates (each 23-29% of conversations): Direct Problem Solving, Direct Output Creation, Collaborative Problem Solving, and Collaborative Output Creation. Students primarily use AI systems for creating (using information to learn something new) and analyzing (taking apart the known and identifying relationships), such as creating coding projects or analyzing law concepts. This aligns with higher-order cognitive functions on Bloom’s Taxonomy . This raises questions about ensuring students don’t offload critical cognitive tasks to AI systems.

Identifying educational AI usage When researching how people use AI models, protecting user privacy is paramount. For this project, we used Claude Insights and Observations, or " Clio ," our automated analysis tool that provides insights into how people are using Claude. Clio enables bottom-up discovery of AI usage patterns by distilling user conversations into high-level usage summaries, such as “troubleshoot code” or “explain economic concepts.” Clio uses a multi-layered, automated process that removes private user information from conversations. We built this process so it minimizes the information that passes from one layer to the next. We describe Clio’s privacy-first design in this earlier blog . We used Clio to analyze approximately one million anonymized 1 conversations from Claude.ai Free and Pro accounts tied to higher education email addresses. 2 We then filtered these conversations for student and academic relevance—such as whether the conversation pertained to coursework or academic research—which yielded 574,740 conversations. 3 Clio then grouped these conversations to derive aggregate education-related insights: how different academic subjects were represented; how students-AI interaction differed; and the types of cognitive tasks that students delegate to AI systems. What are students using AI for? We found that students primarily use Claude to create and improve educational content across disciplines (39.3% of conversations). This often entailed designing practice questions, editing essays, or summarizing academic material. Students also frequently used Claude to provide technical explanations or solutions for academic assignments (33.5%)—working with AI to debug and fix errors in coding assignments, implement programming algorithms and data structures, and explain or solve mathematical problems. Some of this usage might also be cheating, which we discuss below. A smaller but still sizable portion of student usage was to analyze and visualize data (11.0%), support research design and tool development (6.5%), create technical diagrams (3.2%), and translate or proofread content between languages (2.4%). Below is a more detailed breakdown of common requests across subjects. Common student requests from the top four subject areas, based on the 15 most frequent requests in Clio within each subject. AI usage across academic disciplines We next examined which subjects showed disproportionate use of Claude. We did so by comparing Claude.ai usage patterns with the number of U.S. bachelor's degrees awarded. 4 The most disproportionately heavy use of Claude was in Computer Science: despite representing only 5.4% of U.S. bachelor's degrees, Computer Science accounted for 38.6% of conversations on Claude.ai (this might reflect Claude’s particular strengths in computer coding). Natural Sciences and Mathematics also show higher representation in Claude.ai relative to student enrollment (15.2% vs. 9.2%, respectively). Conversely, Business-related educational conversations accounted for just 8.9% of conversations despite constituting 18.6% of bachelor's degrees, showing a disproportionately low use of Claude. Health Professions (5.5% vs. 13.1%) and Humanities (6.4% vs. 12.5%) were also less represented relative to student enrollment in these disciplines. These patterns suggest that STEM students, particularly those in Computer Science, may be earlier adopters of Claude for educational purposes, while students in Business, Health, and Humanities disciplines may be integrating these tools more slowly into their academic workflows. This may reflect higher awareness of Claude in Computer Science communities, as well as AI systems’ greater proficiency at tasks performed by STEM students relative to those performed by students in other disciplines. Comparing the percentage of Claude.ai student conversations that are related to an National Center for Education Statistics ( NCES ) subject area (gray) to the percentage of U.S. college students with an associated major (orange). Note that percentages don’t sum to 100% as some conversations were classified under the “Other” category from the NCES which we exclude from our analysis. How students interact with AI There are many ways of interacting with AI, and they’ll affect the learning process differently. In our analysis of how students interact with AI, we identified four distinct patterns of interaction, which we categorized along two different axes, as shown in the figure below. The first axis was “mode of interaction”. This could involve: 5 (1) Direct conversations, where the user is looking…