Study tests whether AI can compose beautiful chess problems

DeepMind tested if models compose chess puzzles viewed as beautiful. Expert judges rated some AI works competitive, prompting debates on creativity metrics, authorship, and disclosure.