Mitigating Social Desirability Bias in Random Silicon Sampling

Sashank Chapala, Maksym Mironov, Songgaojun Deng

Large Language Models (LLMs) are increasingly used to simulate population responses, a method known as ``Silicon Sampling''. However, responses to socially sensitive questions freq…