twitter thread on paper
Model suggests 3 things we need to measure to know impact of AI on science: - share of research tasks that AI can perform- performance of AI on tasks- strength of bottlenecks (arithmetic mean vs harmonic mean)
Repost this lynk