What did Anthropic analyze?
Anthropic says it classified 207,049 consumer-plan Claude and Claude Code sessions from Aug. 2 through Sept. 30, 2026, assigning them to scientific fields and research tasks.
Answered in
Anthropic Counted 207,000 Science Sessions. What the Count Can’t ProveAnthropic’s study maps Claude sessions to scientific fields and tasks. It describes product use, not a causal measure of research progress.
Read the full analysisOther questions this article answers
More ai research questions
- Does Anthropic’s study show that AI models are conscious?
- Why could cooperation between agents create a safety risk?
- What should developers take from the findings?
- What is BioBasicsBench?
- Does the benchmark show that an AI can run a laboratory?
- How should readers interpret the reported score?
- Does the study show that Claude speeds up scientific discovery?
- Are the findings representative of all scientists?
Every answer on Crashtech is written by the editor of the article it comes from — never auto-summarised. Browse all answers or the Ai Research beat.