Anthropic Education Report: How university students use Claude

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.

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 .

Our research is grounded in real-world data. That has many advantages in terms of the validity of our findings and their application to educational contexts. However, it also comes with limitations that might affect the scope of our findings:

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%).

The fact that AI systems exhibit these skills does not preclude students from also engaging in the skills themselves—for example, co-creating a project together or using AI-generated code to analyze a dataset in another context—but it does point to the potential concerns of students outsourcing cognitive abilities to AI. There are legitimate worries that AI systems may provide a crutch for students, stifling the development of foundational skills needed to support higher-order thinking. An inverted pyramid, after all, can topple over.

We've seen in our discussions with students and educators that AI can empower learning in remarkable ways. For example, AI has been used to support a student's nuclear fusion reactor project , and to facilitate better communication between students and teachers in classrooms .

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.

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.