A report published by Anthropic on November 5, 2026, provides an in-depth analysis of how scientists are leveraging the Claude AI platform to enhance research workflows. The study, based on 207,049 research sessions conducted between August 2 and September 30, 2026, examines the adoption of Claude, Claude Code, and the recently introduced Claude Science across various scientific fields and tasks. The findings highlight the potential of AI to accelerate scientific discovery, particularly in computationally intensive disciplines.
AI Adoption Varies Across Fields and Tasks
The report reveals that AI adoption is more prevalent in formal sciences such as computer science, mathematics, and statistics, relative to the number of researchers in these fields. In contrast, health sciences and lab-based physical sciences exhibit lower levels of AI use. Data analysis emerges as the most common application, accounting for approximately one-third of Claude sessions and nearly half of Claude Code sessions. Writing is another significant use case, while tasks like idea generation and data gathering are less frequently performed, despite being rated as critical for discovery in an accompanying survey of 528 researchers.
Notably, the specialized Claude Science platform, launched in beta on October 14, 2026, shows a higher share of sessions dedicated to idea generation, suggesting that tailored AI tools can address gaps in current usage patterns. The platform integrates advanced features such as protein and chemical structure rendering, code-linked reproducible outputs, and compatibility with high-performance computing environments.
AI Capabilities and Research Acceleration
The study introduces a theoretical measure of AI capability, finding that tasks where AI performs well receive two to four times more usage than tasks where AI is less effective. Fields with a high concentration of cognitive tasks, such as computer science and statistics, benefit the most from AI adoption. In a hypothetical scenario where AI fully automates all non-physical research tasks, output in these fields could increase up to 15-fold, while bench sciences like chemistry and biology could see a three- to fourfold rise due to physical bottlenecks.
Surveyed researchers express optimism about AI's future capabilities, with 40% expecting AI to independently conduct full research projects by 2030. However, the report also cautions about potential risks, including reduced human oversight and ethical concerns, as well as the broader impact on researchers' careers and the nature of scientific work.
Implications for the Future of Science
The findings underscore the transformative potential of AI in scientific research, particularly in computational and cognitive domains. By focusing on areas where AI excels, such as data analysis and writing, researchers can significantly enhance productivity. The development of specialized platforms like Claude Science may further expand AI's utility, addressing critical tasks like idea generation and hypothesis formation. However, the report emphasizes the need for careful consideration of the ethical and practical implications of AI-driven research.
