Simulating Identity, Propagating Bias: Abstraction and Stereotypes in LLM-Generated Text

Open in new window