Gender bias across LLMs is common and highly heterogeneous

Edoardo Bolzoni, Valerio Capraro

Understanding gender biases in large language models (LLMs) is increasingly important as these systems become embedded in decision-support tools with real consequences. Prior resea…