Reinforcing Agentic Creativity in Scientific Ideation with Night Science

Priyanka Kargupta, Silviu Cucerzan, Shweti Mahajan +4

Large language models (LLMs) excel at structured, verifiable tasks, but their low-entropy bias can produce homogeneous and predictable outputs, limiting their utility for open-ende…