LLM Bias and the Original Position for Algorithmic Fairness

This analysis examines how LLM parameter encodings of linguistic representations influence agentic fairness. It evaluates whether models can simulate Rawlsian justice by decoupling decision-making from encoded social identities like race, gender, or nationality.