A synthetic dataset of 30 paired persona profiles contrasting socioeconomically advantaged and disadvantaged versions of the same archetypal character. Each persona spans four life dimensions, making the dataset useful for fairness research, bias evaluation, and controlled preference studies in LLM settings.
Each row represents one persona archetype (e.g., "AI researcher", "real estate agent", "police officer")… See the full description on the dataset page:
https://huggingface.co/datasets/groupfairnessllm/persona.
