Tag
The paper examines how inference setup shapes large language model behavior in medical resource allocation, showing context-dependent biases and emphasizing the importance of careful integration into decision-making systems.