Investigating Gender Stereotypes in Large Language Models via Social Determinants of Health

A new research study reveals that Large Language Models (LLMs) propagate gender stereotypes and biases when processing healthcare data, particularly through interactions between gender and social determinants of health. The research used French patient records to demonstrate how LLMs rely on embedded stereotypes to make gendered decisions in healthcare contexts.

• →LLMs perpetuate gender biases and stereotypes embedded in their training data, particularly in sensitive healthcare applications.
• →Current bias evaluation methods often miss important interactions between different social determinants of health factors.
• →The study used French patient records to probe relationships between gender and other social health factors.
• →LLMs make gendered decisions based on embedded stereotypes rather than objective medical information.
• →Evaluating interactions among social determinants of health could improve bias assessment in AI models.