AI CURATED HOT STORY · 11/8/2026
Anthropic released a report based on approximately 207,000 research-related Claude sessions and a survey of 528 scientists, revealing a significant gap between AI's actual use in research and scientists' expectations. Data analysis is the top use case, accounting for 33.7% of Claude research sessions and 48.1% of Claude Code sessions, followed by writing at 32%. However, hypothesis generation, rated as most important by scientists, represents only 4.2% of sessions, showing both capability and willingness constraints.
The report quantifies task alignment using an 'AI capability score' and 'usage multiplier': tasks in the top 20% of capability are used 2 to 4 times more, while the bottom 20% are almost unused. Computer science, math, and statistics show the highest usage density; AI/ML researchers, though only 1.2% of global researchers, contribute 15.2% of Claude Code research sessions, indicating that task AI-ability drives adoption.
Anthropic estimates that if AI handled all non-manual work, research output in computer science and statistics could accelerate by about 15 times, while chemistry and biology, limited by experiments, would see only 3 to 4 times. This gap highlights that AI's impact on research is highly discipline-dependent, with experimental sciences still bottlenecked by physical operations, while quantitative fields may achieve efficiency leaps first.
The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.