Using LLMs to judge the novelty of AI scientists’ ideas

A post recommends work led by the author’s student Noy to people building AI scientists and using LLMs to assess whether those systems’ ideas are novel. The author calls it “great work.”