Anthropic research reveals scientists primarily use Claude for data analysis and writing

A recent report from Anthropic indicates that researchers predominantly utilize the Claude AI for tasks involving data analysis and writing, while less frequently employing it for data collection or generating research hypotheses. The study analyzed 207,000 research conversations between August 2 and September 30, 2026, comprising 129,000 Claude dialogues and 78,000 Claude Code sessions. Discipline-wise, computer science, mathematics, and statistics showed the highest usage rates, with computer science accounting for 11.9% of Claude conversations. Furthermore, Claude Code usage was heavily concentrated in quantitative fields; for instance, CS researchers made up only 4.3% of the global researcher pool but contributed 32.7% of Claude Code sessions. Data analysis dominated Claude conversations at 33.7%, compared to 27.6% of researchers' actual time spent on that task. Anthropic noted that usage correlates with task suitability across disciplines. A survey of 528 active AI users, primarily from life and health sciences, corroborated the findings, showing that data analysis and writing were the most common AI applications among scientists.