Press ESC to close

How Scientists Use AI: Anthropic Analyses 207,000 Claude Research Sessions

Artificial intelligence is becoming an increasingly useful tool in scientific research. Researchers can use AI to analyse data, write research documents, develop software, and explore scientific information.

But how are scientists actually using AI today?

On 5 November 2026, Anthropic published a report titled AI for Scientific Discovery: How Researchers Use Claude. The study analysed more than 207,000 research sessions to understand how AI is being used across scientific fields and research activities.

What Did Anthropic Study?

Anthropic analysed 207,049 research sessions from 2 August to 30 September 2026.

The main dataset included:

  • 128,914 sessions from Claude's general-purpose interfaces.
  • 78,135 sessions from Claude Code.
  • An additional sample of Claude Science sessions.

The research team used privacy-preserving classification to identify scientific fields and research tasks. The study retained task labels rather than the underlying research conversations or unpublished scientific findings.

It is important to understand that these were research sessions, not 207,000 individual scientists.

What Do Scientists Use AI For?

One of the most interesting findings is that researchers frequently use AI for data analysis and writing.

The report found that data analysis accounted for 33.7% of Claude research sessions and 48.1% of Claude Code research sessions.

Writing and communication accounted for 32% of Claude sessions.

By comparison, data gathering and idea generation represented smaller shares of usage.

This suggests that AI is currently more commonly used to support existing research workflows than to independently generate entirely new scientific research directions.

Which Scientific Fields Use AI More?

Anthropic found relatively high Claude usage in computational and formal sciences.

These include:

  • Computer science
  • Mathematics
  • Statistics
  • Artificial intelligence and machine learning
  • Computational biology

Computer science accounted for 32.7% of Claude Code research sessions in the study.

Health sciences and laboratory-heavy physical sciences showed lower relative usage. One reason may be that many scientific activities require physical experiments, patient interactions, or specialised laboratory equipment.

What Is Claude Science?

Claude Science is a specialised research interface designed to connect Claude with scientific tools and databases.

For example, a biologist studying a protein may need information about its structure, biological function, related compounds, and previous research papers.

A specialised AI workflow can help connect these different sources and assist with building testable scientific hypotheses.

Anthropic found that hypothesis-generation tasks accounted for 8% of Claude Science sessions, a higher share than in its general-purpose interfaces.

Can AI Accelerate Scientific Discovery?

Anthropic also explored possible future scenarios for scientific research.

Under a scenario where existing AI capabilities are widely adopted, its model estimates that research output could approach twice the current level in some fields, including statistics and computer science.

It also explored a hypothetical future in which AI automates all non-physical research steps.

In that scenario, some computational fields could potentially reach around 15 times the research output. Laboratory-heavy fields such as biology and chemistry could reach approximately three to four times the output because physical experimentation remains a major constraint.

These are illustrative scenarios, not predictions or measured productivity improvements. They depend on assumptions about AI capabilities, costs, adoption, and research workflows.

What Are the Limitations?

The research is based on Claude consumer-plan usage and does not represent every scientist, research institution, or AI platform.

The accompanying survey included 528 scientists, primarily active AI users in life and health sciences.

AI is also less capable of physical tasks such as conducting experiments, collecting laboratory samples, and interacting with patients.

Scientific discovery requires more than analysing data or writing reports. It also requires reliable experiments, strong hypotheses, reproducibility, and expert judgement.

What Does This Mean for the Future?

AI could become an increasingly important research assistant.

Scientists may use AI to reduce routine analysis work, connect research databases, prepare technical documents, and explore scientific possibilities.

However, AI-generated conclusions must be checked. More AI-assisted research activity does not automatically mean more genuine scientific breakthroughs.

Conclusion

Anthropic's analysis of 207,049 Claude research sessions provides an interesting picture of how scientists are using AI today.

Data analysis and writing are already important applications. More specialised research tools may help AI support increasingly complex scientific tasks.

The future of scientific discovery may involve close collaboration between AI systems and human researchers, with physical experiments and scientific validation remaining essential.

For more Tamil explanations about AI and emerging technology, follow Neural AI Mind.

Sources

Leave a comment

Your email address will not be published. Required fields are marked *