This research explores how Large Language Models can transform scientific hypothesis generation by addressing information overload and disciplinary fragmentation.

• Systematically categorizes LLM-based hypothesis generation methods from simple prompting to complex frameworks
• Proposes a taxonomy to organize current approaches in the rapidly evolving field
• Highlights LLMs' potential to integrate cross-disciplinary knowledge and identify novel research directions
• Addresses challenges of evaluating AI-generated scientific hypotheses

For biologists, this research offers practical frameworks to leverage AI as a collaborative tool for uncovering connections in vast research landscapes that might otherwise remain hidden, potentially accelerating discoveries.

A Survey on Hypothesis Generation for Scientific Discovery in the Era of Large Language Models