Researchers analyzed 207,000 Claude research sessions to understand AI's impact on scientific discovery. The study found Claude is used more in formal sciences like computer science and mathematics, with data analysis accounting for about one-third of sessions. Scientists primarily use the tool for analysis and writing, ranking idea generation and data gathering as most important, though they use AI less for these tasks.
The research shows AI could significantly speed up discovery if widely adopted, with existing capabilities potentially doubling research output in some fields. If AI could carry out all non-physical research work, output in computer science and statistics could rise up to 15-fold, while bench sciences like chemistry and biology, limited by physical bottlenecks, would see three- or fourfold increases. A survey of active AI-using researchers found about 40% expect AI models to carry out full research projects independently by 2030.
The study examined usage patterns across scientific fields, revealing that tasks AI is best suited for receive two to four times more usage than predicted, while least suited tasks see minimal use. The findings suggest specialized AI tools could narrow the gap between how scientists currently use AI and what they consider most important for discovery.
Source: Hacker News · Summarized by HeadlinesBriefing